Master Art of Translation | The Implicit Coherence Transfer System — How Unspoken Causes, Contrasts, Concessions and Explanations Keep Their Meaning

Translation coherence: the quick answer

How do you translate implicit discourse relations, connectives and paragraph coherence without adding a cause, contrast or promise that the author never asserted? The reliable method is to map the relationship between adjacent ideas before selecting words such as because, although, therefore, meanwhile, however or for example. An accurate translation preserves both the propositions and the degree to which their relationship is explicit. When the source only places two facts side by side, a polished target sentence must not silently convert their proximity into proof of causation.

The hidden translation problem is not a missing dictionary entry. It is an invisible bridge between clauses: readers infer whether a second sentence explains the first, qualifies it, contradicts an expectation, adds independent information or simply follows it in time. Languages and genres differ in how often they mark those bridges. A translator who automatically inserts a connective can make prose sound smoother while changing its evidential burden, the reader’s interpretation, and sometimes an operational decision. A translator who removes an essential connective may produce equally fluent sentences that no longer tell readers why one claim matters to another.

This complete guide teaches a repeatable system for translating coherence relations, including implicit cause and effect, concession versus contrast, explanations versus justifications, temporal sequence versus result, exception versus qualification, and examples versus evidence. It offers an interpretation map, a step-by-step review procedure, constructed diagnostic cases and transfer exercises for research writing, dialogue, public notices, instruction manuals, journalism and literature. It also shows how to handle genuine ambiguity: preserve an unstated relationship when the evidence is insufficient, or ask the author rather than upgrading a plausible inference into a source claim.

The core invariant is simple: a reader of the translation should infer no stronger, weaker, differently directed or differently attributed relation than a reasonable reader of the source in its intended context. The surface wording may change. The reasoning path must remain defensible. A well-edited target text should sound natural without concealing the difference between what the source says, what it implies, what a translator infers and what a reader is invited to conclude.

• Separate individual propositions from the relation that binds them; one may survive while the other is mistranslated.
• Identify the relation family before choosing a connective: cause, result, purpose, contrast, concession, condition, time, elaboration, explanation or evidence.
• Distinguish explicit relations signalled by words from implicit relations inferred through context and genre.
• Write a neutral meaning brief before drafting polished target sentences.
• Ask whether a candidate connective introduces causality, agency, certainty, exceptions or speaker commitment.
• Try a minimally committed version and a more explicit version; retain the stronger one only when licensed.
• Test reversals: because is not therefore, although is not whereas, after is not as a result.
• Look beyond a single sentence; paragraph-level emphasis and the writer’s reasoning path often determine the correct link.
• Separate a discourse interpretation problem from a simple grammatical or punctuation error.
• Record uncertain bridges as review questions rather than hiding assumptions in fluent prose.
• Check both accuracy and reading effort; needless opacity can also reduce the usefulness of a translation.
• Release only when the relation map, reader action and source evidence agree with the target.

How to use this guide

First read the worked decision method. Then select a relation family that matches the text you are translating. Do the small diagnostic exercises without looking at the suggested repair: name the two propositions, state the relation, compare alternative connectives, and explain which meaning is lost by each alternative. Finally, use the transfer workshops to apply the same method to a different genre. The purpose is not to memorize a list of linking words. It is to learn what each link commits the writer to.

All examples below are newly constructed English-language meaning probes. They model a source-side discourse problem and possible English target-language renderings; they are not quotations from another language or claims that any particular language always expresses a relation in one fixed way. Where the intended meaning comes from a hypothetical author brief, that condition is stated. In real work, context, source grammar, speaker intent, subject expertise and sometimes consultation supply that evidence.

1. What coherence means, and why translators lose it

Two sentences can each be faithfully translated and still form a misleading paragraph. ‘The sample was stored overnight. The result was delayed’ reports two situations. ‘The result was delayed because the sample was stored overnight’ commits to a causal explanation that the unmarked pair alone does not establish. ‘Although the sample was stored overnight, the result was delayed’ suggests an expected benefit from storage and a disappointment. ‘The sample was stored overnight; meanwhile, the result was delayed’ suggests temporal overlap. Each version reuses much of the same vocabulary, but changes the mental model of the events.

A discourse relation is an interpretive connection between spans of language, such as clauses, sentences or larger blocks. It may be indicated by a connective, a grammatical construction, a lexical expression, an ordering convention or shared knowledge. It may also remain unspoken. Coherence is the reader’s ability to construct a sensible connected representation. Cohesion describes surface resources such as pronouns, repetition and linkers that help produce that connection. Cohesion can signal coherence but cannot guarantee it.

The critical distinction is between a relation being inferable and a relation being asserted. A laboratory note may leave a reason unmarked because it is not yet confirmed. An editor can make that note apparently clearer by adding ‘therefore’, but the added word may turn a provisional observation into a claim of demonstrated consequence. A literary narrator may juxtapose two events to let the reader discover an irony. An enthusiastic translator who writes ‘despite this’ can narrow the ambiguity and change the narrator’s voice.

Different discourse theories carve the territory differently. The Penn Discourse Treebank describes families including temporal, contingency, comparison and expansion; Rhetorical Structure Theory models how spans perform roles in larger text structure. These are useful analysis tools, not interchangeable universal taxonomies. A translator can benefit from either without forcing every paragraph into a uniquely correct annotation.

Recent research helps explain why this is a practical problem rather than merely a stylistic preference. Yung and colleagues studied English–German aligned discourse connectives, finding evidence for more explicit discourse relations in translations than in comparable originally written texts. That result should not be turned into a rule that all translation must be explicit, or a claim about every language pair. It is evidence that translators and language systems often alter connective marking in systematic ways, making relation review a worthwhile quality-control task.

The central question is not ‘Which conjunction sounds best?’ It is ‘What change in the reader’s understanding does this conjunction cause?’ Once that question is habitual, many apparently minor errors become visible. The distinction is useful for humans translating by hand, editors revising machine output, reviewers checking subtitles, and multilingual teams maintaining explanations across document versions.

2. The hidden problem: the relation can disappear while the sentences remain grammatical

1. A missing connective is not necessarily a missing meaning

Juxtaposition often allows the reader to infer a link while leaving its strength open. The translator must inspect paragraph goals, earlier evidence, speaker roles and the genre before deciding whether marking the link would preserve clarity or introduce an unsupported commitment. Lack of an overt marker is a fact about the text, not proof that the two events are unrelated.

Applied verification: first isolate the two spans involved in this particular a missing connective is not necessarily a missing meaning problem. State what a reader would believe before and after the proposed connective, including whether the second span supplies a cause, merely follows in time, qualifies a claim or guides attention. Compare a minimally marked translation with an explicit one. Retain the explicit version only if a visible source cue, reliable context or documented author brief supports the extra commitment. An elegant sentence that invents a relationship is not a faithful sentence.

2. An added connective can become a new factual claim

Because, so, therefore and consequently may assert a dependency that a source merely suggests. In a report, that alters an explanatory hypothesis; in instructions, it can alter what is a prerequisite; in public communication, it can assign responsibility. Check whether the information licenses the chosen relation rather than accepting the best-sounding English sentence.

Applied verification: first isolate the two spans involved in this particular an added connective can become a new factual claim problem. State what a reader would believe before and after the proposed connective, including whether the second span supplies a cause, merely follows in time, qualifies a claim or guides attention. Compare a minimally marked translation with an explicit one. Retain the explicit version only if a visible source cue, reliable context or documented author brief supports the extra commitment. An elegant sentence that invents a relationship is not a faithful sentence.

3. Concession predicts an expectation that ordinary contrast need not predict

Although and nevertheless imply that one situation might have been expected to hinder or exclude the other. Whereas and by contrast simply compare. Writing ‘although’ to join two different facts can invent an unspoken social judgement or causal expectation. Review the surrounding argument to decide whether an expectation is part of the source.

Applied verification: first isolate the two spans involved in this particular concession predicts an expectation that ordinary contrast need not predict problem. State what a reader would believe before and after the proposed connective, including whether the second span supplies a cause, merely follows in time, qualifies a claim or guides attention. Compare a minimally marked translation with an explicit one. Retain the explicit version only if a visible source cue, reliable context or documented author brief supports the extra commitment. An elegant sentence that invents a relationship is not a faithful sentence.

4. Chronology and causality must not be collapsed

After states an ordering relation. Therefore states inference or consequence. Two events being adjacent in a narrative does not establish that the first brought about the second. The relation can be tested by a counterfactual: could the second event have happened independently of the first while the original passage remained true?

Applied verification: first isolate the two spans involved in this particular chronology and causality must not be collapsed problem. State what a reader would believe before and after the proposed connective, including whether the second span supplies a cause, merely follows in time, qualifies a claim or guides attention. Compare a minimally marked translation with an explicit one. Retain the explicit version only if a visible source cue, reliable context or documented author brief supports the extra commitment. An elegant sentence that invents a relationship is not a faithful sentence.

5. Explanation and justification have different targets

An explanation answers why an event happened. A justification answers why a conclusion or action should be accepted. ‘The road was closed because the bridge failed’ explains a closure; ‘The closure was sensible because the bridge had failed’ justifies a decision. Both include because, but the target of the connection differs. Translations may wrongly shift that target.

Applied verification: first isolate the two spans involved in this particular explanation and justification have different targets problem. State what a reader would believe before and after the proposed connective, including whether the second span supplies a cause, merely follows in time, qualifies a claim or guides attention. Compare a minimally marked translation with an explicit one. Retain the explicit version only if a visible source cue, reliable context or documented author brief supports the extra commitment. An elegant sentence that invents a relationship is not a faithful sentence.

6. A condition is not an observed outcome

If a sensor exceeds a limit, the system alerts describes a rule; the sensor exceeded a limit, so the system alerted describes a reported event and its consequence. Substituting one for the other changes a future instruction into a historical statement, or vice versa. Check mood, tense, genericity and procedural context together.

Applied verification: first isolate the two spans involved in this particular a condition is not an observed outcome problem. State what a reader would believe before and after the proposed connective, including whether the second span supplies a cause, merely follows in time, qualifies a claim or guides attention. Compare a minimally marked translation with an explicit one. Retain the explicit version only if a visible source cue, reliable context or documented author brief supports the extra commitment. An elegant sentence that invents a relationship is not a faithful sentence.

7. Contrast is not contradiction

One speaker can describe a lower cost and another describe a slower delivery without either claim refuting the other. However may invite the reader to balance competing dimensions, but contradicts is a stronger claim. A translator must distinguish comparison between dimensions from opposition of truth values.

Applied verification: first isolate the two spans involved in this particular contrast is not contradiction problem. State what a reader would believe before and after the proposed connective, including whether the second span supplies a cause, merely follows in time, qualifies a claim or guides attention. Compare a minimally marked translation with an explicit one. Retain the explicit version only if a visible source cue, reliable context or documented author brief supports the extra commitment. An elegant sentence that invents a relationship is not a faithful sentence.

8. Evidence is not identical to the claim it supports

‘Measurements increased; the model may be incomplete’ is a suggestion for inquiry, not a proof that the model has failed. Rendering the first clause as conclusive evidence can erase hedge words and the inferential gap. Audit what is observed separately from what is inferred.

Applied verification: first isolate the two spans involved in this particular evidence is not identical to the claim it supports problem. State what a reader would believe before and after the proposed connective, including whether the second span supplies a cause, merely follows in time, qualifies a claim or guides attention. Compare a minimally marked translation with an explicit one. Retain the explicit version only if a visible source cue, reliable context or documented author brief supports the extra commitment. An elegant sentence that invents a relationship is not a faithful sentence.

9. An example is not a definition

A sentence beginning for example illustrates a category without exhausting it. ‘Such as’ can create an open class; ‘namely’ may introduce an exhaustive or specifying list depending on context. Translators should test whether readers treat the examples as optional instances or the full membership of a group.

Applied verification: first isolate the two spans involved in this particular an example is not a definition problem. State what a reader would believe before and after the proposed connective, including whether the second span supplies a cause, merely follows in time, qualifies a claim or guides attention. Compare a minimally marked translation with an explicit one. Retain the explicit version only if a visible source cue, reliable context or documented author brief supports the extra commitment. An elegant sentence that invents a relationship is not a faithful sentence.

10. A restatement should not smuggle in new conditions

In other words promises essentially the same proposition, perhaps in more accessible language. If the second sentence adds a new threshold, exception or limitation, it is not merely a paraphrase. A supposedly clarifying rewrite can be more restrictive than the source and therefore require an explicit qualifier.

Applied verification: first isolate the two spans involved in this particular a restatement should not smuggle in new conditions problem. State what a reader would believe before and after the proposed connective, including whether the second span supplies a cause, merely follows in time, qualifies a claim or guides attention. Compare a minimally marked translation with an explicit one. Retain the explicit version only if a visible source cue, reliable context or documented author brief supports the extra commitment. An elegant sentence that invents a relationship is not a faithful sentence.

11. Background is not a subordinate fact by default

A short historical detail can be indispensable for understanding a claim without being its cause. Translating it as the reason for a present event can distort a writer’s structure. The reviewer should identify whether the span answers when, why, what context, or what evidence makes the central point intelligible.

Applied verification: first isolate the two spans involved in this particular background is not a subordinate fact by default problem. State what a reader would believe before and after the proposed connective, including whether the second span supplies a cause, merely follows in time, qualifies a claim or guides attention. Compare a minimally marked translation with an explicit one. Retain the explicit version only if a visible source cue, reliable context or documented author brief supports the extra commitment. An elegant sentence that invents a relationship is not a faithful sentence.

12. A list’s relationship can change across paragraphs

First, second and finally can count items, steps, priorities, time periods or rhetorical points. Changing a list of options into chronological instructions can change what a reader does. Where the source leaves order nonbinding, target words such as next or then may accidentally impose a sequence.

Applied verification: first isolate the two spans involved in this particular a list’s relationship can change across paragraphs problem. State what a reader would believe before and after the proposed connective, including whether the second span supplies a cause, merely follows in time, qualifies a claim or guides attention. Compare a minimally marked translation with an explicit one. Retain the explicit version only if a visible source cue, reliable context or documented author brief supports the extra commitment. An elegant sentence that invents a relationship is not a faithful sentence.

13. The word while has multiple relation functions

While can mark simultaneity or contrast, and the appropriate reading comes from semantics and context. ‘While the pump runs, record the pressure’ is temporal; ‘While the pump is inexpensive, it is noisy’ is concessive. A translation that picks one automatically can invert procedural meaning.

Applied verification: first isolate the two spans involved in this particular the word while has multiple relation functions problem. State what a reader would believe before and after the proposed connective, including whether the second span supplies a cause, merely follows in time, qualifies a claim or guides attention. Compare a minimally marked translation with an explicit one. Retain the explicit version only if a visible source cue, reliable context or documented author brief supports the extra commitment. An elegant sentence that invents a relationship is not a faithful sentence.

14. Yet, still and even so carry stance

These expressions do more than join clauses. They indicate that a counter-expectation survives an earlier fact. If the source simply adds another property, retaining a strong concessive marker may overstate surprise. If the source relies on surprise, omitting the marker can flatten an author’s deliberate contrast.

Applied verification: first isolate the two spans involved in this particular yet, still and even so carry stance problem. State what a reader would believe before and after the proposed connective, including whether the second span supplies a cause, merely follows in time, qualifies a claim or guides attention. Compare a minimally marked translation with an explicit one. Retain the explicit version only if a visible source cue, reliable context or documented author brief supports the extra commitment. An elegant sentence that invents a relationship is not a faithful sentence.

15. Lexical signals can encode links without conjunctions

An expression such as ‘this explains the delay’ or ‘the consequence was visible’ supplies a relation even without because or so. Translators who look only for traditional connectives will miss these signals. Audit nouns and verbs that encode result, reason, exception, proof, contrast and purpose.

Applied verification: first isolate the two spans involved in this particular lexical signals can encode links without conjunctions problem. State what a reader would believe before and after the proposed connective, including whether the second span supplies a cause, merely follows in time, qualifies a claim or guides attention. Compare a minimally marked translation with an explicit one. Retain the explicit version only if a visible source cue, reliable context or documented author brief supports the extra commitment. An elegant sentence that invents a relationship is not a faithful sentence.

16. Relation scope can extend beyond adjacent sentences

A however near the start of a paragraph may contrast the whole preceding discussion, not merely the immediately previous sentence. A reason introduced in one clause may license several subsequent recommendations. Annotate spans rather than pairing every connective with the closest visible sentence.

Applied verification: first isolate the two spans involved in this particular relation scope can extend beyond adjacent sentences problem. State what a reader would believe before and after the proposed connective, including whether the second span supplies a cause, merely follows in time, qualifies a claim or guides attention. Compare a minimally marked translation with an explicit one. Retain the explicit version only if a visible source cue, reliable context or documented author brief supports the extra commitment. An elegant sentence that invents a relationship is not a faithful sentence.

17. The reader’s default inference may be culture- or genre-dependent

Familiar text conventions differ, but it is risky to invent essentializing rules about what entire language communities assume. Analyze the intended audience and the actual document rather than treating culture as a fixed switch. When ambiguity matters, prefer evidence from readers in the target context.

Applied verification: first isolate the two spans involved in this particular the reader’s default inference may be culture- or genre-dependent problem. State what a reader would believe before and after the proposed connective, including whether the second span supplies a cause, merely follows in time, qualifies a claim or guides attention. Compare a minimally marked translation with an explicit one. Retain the explicit version only if a visible source cue, reliable context or documented author brief supports the extra commitment. An elegant sentence that invents a relationship is not a faithful sentence.

18. The source author may have intended productive ambiguity

Literary writing, interviews and exploratory essays sometimes leave two relationships simultaneously plausible. A single explanatory connective can destroy that openness. In such cases, an unmarked translation, a punctuation choice or an alternative construction may better protect the text’s interpretive range.

Applied verification: first isolate the two spans involved in this particular the source author may have intended productive ambiguity problem. State what a reader would believe before and after the proposed connective, including whether the second span supplies a cause, merely follows in time, qualifies a claim or guides attention. Compare a minimally marked translation with an explicit one. Retain the explicit version only if a visible source cue, reliable context or documented author brief supports the extra commitment. An elegant sentence that invents a relationship is not a faithful sentence.

19. Genre changes the cost of the same connective error

A false ‘therefore’ in a poem may redirect interpretation; in a safety instruction it may cause a wrong action; in a report it may misstate causal evidence. Severity depends on purpose, reader reliance and consequence, not on whether the target sentence sounds awkward.

Applied verification: first isolate the two spans involved in this particular genre changes the cost of the same connective error problem. State what a reader would believe before and after the proposed connective, including whether the second span supplies a cause, merely follows in time, qualifies a claim or guides attention. Compare a minimally marked translation with an explicit one. Retain the explicit version only if a visible source cue, reliable context or documented author brief supports the extra commitment. An elegant sentence that invents a relationship is not a faithful sentence.

20. Review must distinguish explanation from blame

A message may report that a service was interrupted after maintenance. Writing because of maintenance can assign causality; writing due to operator error can assign blame. If the source does not make that claim, the translation must not manufacture an accountable agent.

Applied verification: first isolate the two spans involved in this particular review must distinguish explanation from blame problem. State what a reader would believe before and after the proposed connective, including whether the second span supplies a cause, merely follows in time, qualifies a claim or guides attention. Compare a minimally marked translation with an explicit one. Retain the explicit version only if a visible source cue, reliable context or documented author brief supports the extra commitment. An elegant sentence that invents a relationship is not a faithful sentence.

21. Multi-clause sentences may contain nested relations

A paragraph can state a condition within a concession within a broader argument. Flattening this into a chain of because and so markers can alter which condition controls which outcome. Draw the relation tree first; convert it to fluent syntax only after identifying nested spans.

Applied verification: first isolate the two spans involved in this particular multi-clause sentences may contain nested relations problem. State what a reader would believe before and after the proposed connective, including whether the second span supplies a cause, merely follows in time, qualifies a claim or guides attention. Compare a minimally marked translation with an explicit one. Retain the explicit version only if a visible source cue, reliable context or documented author brief supports the extra commitment. An elegant sentence that invents a relationship is not a faithful sentence.

22. Over-explicitation and under-explicitation are both errors when they change function

Adding an unnecessary linker can overcommit; removing a necessary qualifier can under-specify. The goal is neither maximum literalness nor maximum smoothness. It is the most natural target structure consistent with the source’s obligations, reader tasks and supported inferences.

Applied verification: first isolate the two spans involved in this particular over-explicitation and under-explicitation are both errors when they change function problem. State what a reader would believe before and after the proposed connective, including whether the second span supplies a cause, merely follows in time, qualifies a claim or guides attention. Compare a minimally marked translation with an explicit one. Retain the explicit version only if a visible source cue, reliable context or documented author brief supports the extra commitment. An elegant sentence that invents a relationship is not a faithful sentence.

23. Discourse relation choice must survive back-translation tests

A helpful adversarial test is to paraphrase the target into a simple pair of propositions and the relation it asserts. Ask whether that pair could be read back as the source meaning without adding premises. This is a diagnostic, not a guarantee: back-translation can reproduce an error if reviewers share the same assumption.

Applied verification: first isolate the two spans involved in this particular discourse relation choice must survive back-translation tests problem. State what a reader would believe before and after the proposed connective, including whether the second span supplies a cause, merely follows in time, qualifies a claim or guides attention. Compare a minimally marked translation with an explicit one. Retain the explicit version only if a visible source cue, reliable context or documented author brief supports the extra commitment. An elegant sentence that invents a relationship is not a faithful sentence.

24. Human review is particularly important when the sentence is too plausible

Fluent translation can hide a relation error precisely because a reader easily supplies a narrative. Reviewers should temporarily forget whether the story ‘makes sense’ and demand source evidence for each assertion of cause, concession, priority, dependency or conclusion.

Applied verification: first isolate the two spans involved in this particular human review is particularly important when the sentence is too plausible problem. State what a reader would believe before and after the proposed connective, including whether the second span supplies a cause, merely follows in time, qualifies a claim or guides attention. Compare a minimally marked translation with an explicit one. Retain the explicit version only if a visible source cue, reliable context or documented author brief supports the extra commitment. An elegant sentence that invents a relationship is not a faithful sentence.

25. Documentation creates honest uncertainty

If the writer’s relation cannot be established, record an unresolved query with the source span, competing interpretations and impact. A translation team should not disguise unknown information as a confident connective. High-risk ambiguity needs authoritative resolution; low-risk ambiguity may be preserved.

Applied verification: first isolate the two spans involved in this particular documentation creates honest uncertainty problem. State what a reader would believe before and after the proposed connective, including whether the second span supplies a cause, merely follows in time, qualifies a claim or guides attention. Compare a minimally marked translation with an explicit one. Retain the explicit version only if a visible source cue, reliable context or documented author brief supports the extra commitment. An elegant sentence that invents a relationship is not a faithful sentence.

26. The local goal may be information management rather than logic

Meanwhile, anyway, now and in any case sometimes guide the reader or listener through a conversation rather than assert a strict propositional relation. Treating such markers as formal logical operators can misrepresent tone and the status of earlier material. Check their discourse-management role.

Applied verification: first isolate the two spans involved in this particular the local goal may be information management rather than logic problem. State what a reader would believe before and after the proposed connective, including whether the second span supplies a cause, merely follows in time, qualifies a claim or guides attention. Compare a minimally marked translation with an explicit one. Retain the explicit version only if a visible source cue, reliable context or documented author brief supports the extra commitment. An elegant sentence that invents a relationship is not a faithful sentence.

27. A conditional consequence is not a guaranteed outcome

‘If the form is accepted, processing begins’ names a triggering event, not a promise that acceptance will occur. Translating the conditional as ‘when the form is accepted’ may presuppose a future acceptance. This distinction matters in policies, notices and promises.

Applied verification: first isolate the two spans involved in this particular a conditional consequence is not a guaranteed outcome problem. State what a reader would believe before and after the proposed connective, including whether the second span supplies a cause, merely follows in time, qualifies a claim or guides attention. Compare a minimally marked translation with an explicit one. Retain the explicit version only if a visible source cue, reliable context or documented author brief supports the extra commitment. An elegant sentence that invents a relationship is not a faithful sentence.

28. The presence of a marker does not eliminate polysemy

Because, since, as, while and so can carry more than one function in different contexts. Lexical matching across languages must be checked against clause type, information structure and audience. Treat the connective as a clue, not as a complete semantic annotation.

Applied verification: first isolate the two spans involved in this particular the presence of a marker does not eliminate polysemy problem. State what a reader would believe before and after the proposed connective, including whether the second span supplies a cause, merely follows in time, qualifies a claim or guides attention. Compare a minimally marked translation with an explicit one. Retain the explicit version only if a visible source cue, reliable context or documented author brief supports the extra commitment. An elegant sentence that invents a relationship is not a faithful sentence.

29. An implicit relation can be made explicit by restructuring, not only by adding a word

A heading, punctuation break, clause order or topic sentence can clarify a reader’s path without inserting because or despite. This is especially useful when target-language conventions require clearer grouping but the source does not authorize a stronger claim.

Applied verification: first isolate the two spans involved in this particular an implicit relation can be made explicit by restructuring, not only by adding a word problem. State what a reader would believe before and after the proposed connective, including whether the second span supplies a cause, merely follows in time, qualifies a claim or guides attention. Compare a minimally marked translation with an explicit one. Retain the explicit version only if a visible source cue, reliable context or documented author brief supports the extra commitment. An elegant sentence that invents a relationship is not a faithful sentence.

30. An explanatory translation may need a reviewer note rather than in-text invention

When target readers need specialized background that source readers were expected to have, a separately identified localization note can be safer than an unsupported explanatory sentence inside the translation. Whether such notes are permitted depends on the brief, format and rights.

Applied verification: first isolate the two spans involved in this particular an explanatory translation may need a reviewer note rather than in-text invention problem. State what a reader would believe before and after the proposed connective, including whether the second span supplies a cause, merely follows in time, qualifies a claim or guides attention. Compare a minimally marked translation with an explicit one. Retain the explicit version only if a visible source cue, reliable context or documented author brief supports the extra commitment. An elegant sentence that invents a relationship is not a faithful sentence.

31. Comprehension is an empirical checkpoint

After a translation is edited, give a reader a simple task: identify the cause, identify the condition, or decide whether the two events are only sequential. If readers routinely infer a claim the source does not support, the target still needs work, even if its grammar and terminology are spotless.

Applied verification: first isolate the two spans involved in this particular comprehension is an empirical checkpoint problem. State what a reader would believe before and after the proposed connective, including whether the second span supplies a cause, merely follows in time, qualifies a claim or guides attention. Compare a minimally marked translation with an explicit one. Retain the explicit version only if a visible source cue, reliable context or documented author brief supports the extra commitment. An elegant sentence that invents a relationship is not a faithful sentence.

32. The aim is to preserve the writer’s reasoning topology

A full document forms a network of claims, examples, background, evidence, qualifications and conclusions. The translator should protect the direction and strength of those links. Vocabulary fidelity alone cannot repair a broken relation network, and a relation network cannot excuse incorrect factual terms.

Applied verification: first isolate the two spans involved in this particular the aim is to preserve the writer’s reasoning topology problem. State what a reader would believe before and after the proposed connective, including whether the second span supplies a cause, merely follows in time, qualifies a claim or guides attention. Compare a minimally marked translation with an explicit one. Retain the explicit version only if a visible source cue, reliable context or documented author brief supports the extra commitment. An elegant sentence that invents a relationship is not a faithful sentence.

3. The 28-pass coherence-transfer workflow

Pass 1: Confirm the purpose

Ask who will read the translation, why, and what action or understanding it should support. A literary reader can tolerate deliberate ambiguity that a procedure user cannot; neither audience authorizes invented content.

Practical record for pass 1: keep an evidence line, a rejected alternative, and a reader-effect note. The evidence line should point to the precise source words, earlier context or approved brief that supports the decision. The rejected alternative should explain which additional premise it would require. The reader-effect note should state how the choice would change comprehension or action. This short record makes the decision reviewable across languages, editors and document versions.

Pass 2: Recover document context

Read the title, headings, preceding sentences, following sentences, speaker turns and relevant visuals. A sentence pulled into a translation memory may have lost the very material that disambiguates its relation.

Practical record for pass 2: keep an evidence line, a rejected alternative, and a reader-effect note. The evidence line should point to the precise source words, earlier context or approved brief that supports the decision. The rejected alternative should explain which additional premise it would require. The reader-effect note should state how the choice would change comprehension or action. This short record makes the decision reviewable across languages, editors and document versions.

Pass 3: Segment the argument

Mark propositions as minimal spans, then identify whether each span contains a factual assertion, proposal, example, observation, question or conditional proposition. Do not add linkers yet.

Practical record for pass 3: keep an evidence line, a rejected alternative, and a reader-effect note. The evidence line should point to the precise source words, earlier context or approved brief that supports the decision. The rejected alternative should explain which additional premise it would require. The reader-effect note should state how the choice would change comprehension or action. This short record makes the decision reviewable across languages, editors and document versions.

Pass 4: Preserve proposition identity

Write each proposition in neutral language before interpreting its link. If a sentence says an event was reported, do not rewrite it as an event known to have happened.

Practical record for pass 4: keep an evidence line, a rejected alternative, and a reader-effect note. The evidence line should point to the precise source words, earlier context or approved brief that supports the decision. The rejected alternative should explain which additional premise it would require. The reader-effect note should state how the choice would change comprehension or action. This short record makes the decision reviewable across languages, editors and document versions.

Pass 5: List candidate relations

Consider time, cause, purpose, result, condition, concession, contrast, evidence, justification, elaboration, example and list. Record more than one candidate if the source allows them.

Practical record for pass 5: keep an evidence line, a rejected alternative, and a reader-effect note. The evidence line should point to the precise source words, earlier context or approved brief that supports the decision. The rejected alternative should explain which additional premise it would require. The reader-effect note should state how the choice would change comprehension or action. This short record makes the decision reviewable across languages, editors and document versions.

Pass 6: Locate explicit signals

Underline source connectives and lexical cues, including relation nouns and verbs such as reason, consequence, suggests, for instance and nevertheless. Record their scope.

Practical record for pass 6: keep an evidence line, a rejected alternative, and a reader-effect note. The evidence line should point to the precise source words, earlier context or approved brief that supports the decision. The rejected alternative should explain which additional premise it would require. The reader-effect note should state how the choice would change comprehension or action. This short record makes the decision reviewable across languages, editors and document versions.

Pass 7: Inspect implicit clues

Look for expectations created by tense, semantic roles, conventional order, paragraph goals and prior information. A plausible world story is weaker evidence than a linguistically encoded commitment.

Practical record for pass 7: keep an evidence line, a rejected alternative, and a reader-effect note. The evidence line should point to the precise source words, earlier context or approved brief that supports the decision. The rejected alternative should explain which additional premise it would require. The reader-effect note should state how the choice would change comprehension or action. This short record makes the decision reviewable across languages, editors and document versions.

Pass 8: Determine direction

Distinguish A causes B from B explains A; distinguish A supports B from B justifies A. The same propositions with reversed links can tell a different story.

Practical record for pass 8: keep an evidence line, a rejected alternative, and a reader-effect note. The evidence line should point to the precise source words, earlier context or approved brief that supports the decision. The rejected alternative should explain which additional premise it would require. The reader-effect note should state how the choice would change comprehension or action. This short record makes the decision reviewable across languages, editors and document versions.

Pass 9: Assess strength

Decide whether the source asserts a fact, makes a likely inference, raises a possibility, reports a third-party explanation or preserves uncertainty. A relation can be present while its certainty is limited.

Practical record for pass 9: keep an evidence line, a rejected alternative, and a reader-effect note. The evidence line should point to the precise source words, earlier context or approved brief that supports the decision. The rejected alternative should explain which additional premise it would require. The reader-effect note should state how the choice would change comprehension or action. This short record makes the decision reviewable across languages, editors and document versions.

Pass 10: Identify contrast type

Test whether two facts are different, incompatible, surprising together or offered as alternatives. Pick a target marker only after deciding which reading is licensed.

Practical record for pass 10: keep an evidence line, a rejected alternative, and a reader-effect note. The evidence line should point to the precise source words, earlier context or approved brief that supports the decision. The rejected alternative should explain which additional premise it would require. The reader-effect note should state how the choice would change comprehension or action. This short record makes the decision reviewable across languages, editors and document versions.

Pass 11: Identify time versus dependence

Ask whether the second event could occur without the first. If yes, the source may still be causal, but sequence alone cannot establish that claim.

Practical record for pass 11: keep an evidence line, a rejected alternative, and a reader-effect note. The evidence line should point to the precise source words, earlier context or approved brief that supports the decision. The rejected alternative should explain which additional premise it would require. The reader-effect note should state how the choice would change comprehension or action. This short record makes the decision reviewable across languages, editors and document versions.

Pass 12: Check nested scope

Draw brackets when an if, even though, only if or unless governs several clauses. A modifier may attach to one local action or a whole policy sentence.

Practical record for pass 12: keep an evidence line, a rejected alternative, and a reader-effect note. The evidence line should point to the precise source words, earlier context or approved brief that supports the decision. The rejected alternative should explain which additional premise it would require. The reader-effect note should state how the choice would change comprehension or action. This short record makes the decision reviewable across languages, editors and document versions.

Pass 13: Check example status

Confirm whether listed items are instances, required steps, evidence points, exceptions or the full set. Linkers like for instance and namely have different effects.

Practical record for pass 13: keep an evidence line, a rejected alternative, and a reader-effect note. The evidence line should point to the precise source words, earlier context or approved brief that supports the decision. The rejected alternative should explain which additional premise it would require. The reader-effect note should state how the choice would change comprehension or action. This short record makes the decision reviewable across languages, editors and document versions.

Pass 14: Protect the narrator and speaker

Separate speaker beliefs from narrator facts and editor explanations. Do not make a reported reason look like a verified reason.

Practical record for pass 14: keep an evidence line, a rejected alternative, and a reader-effect note. The evidence line should point to the precise source words, earlier context or approved brief that supports the decision. The rejected alternative should explain which additional premise it would require. The reader-effect note should state how the choice would change comprehension or action. This short record makes the decision reviewable across languages, editors and document versions.

Pass 15: Draft the least committed accurate version

Make a natural target sentence that does not add an unsupported relation. Neutral coordination or punctuation may preserve interpretive latitude.

Practical record for pass 15: keep an evidence line, a rejected alternative, and a reader-effect note. The evidence line should point to the precise source words, earlier context or approved brief that supports the decision. The rejected alternative should explain which additional premise it would require. The reader-effect note should state how the choice would change comprehension or action. This short record makes the decision reviewable across languages, editors and document versions.

Pass 16: Draft the most explicit licensed version

Try a version with a connective whose specific meaning is actually supported. This creates a useful contrast for editing, not a rule to prefer explicit prose.

Practical record for pass 16: keep an evidence line, a rejected alternative, and a reader-effect note. The evidence line should point to the precise source words, earlier context or approved brief that supports the decision. The rejected alternative should explain which additional premise it would require. The reader-effect note should state how the choice would change comprehension or action. This short record makes the decision reviewable across languages, editors and document versions.

Pass 17: Compare truth conditions

Ask what would have to be true for each target version to be correct. If the explicit version requires an extra causal premise, it is not merely stylistic.

Practical record for pass 17: keep an evidence line, a rejected alternative, and a reader-effect note. The evidence line should point to the precise source words, earlier context or approved brief that supports the decision. The rejected alternative should explain which additional premise it would require. The reader-effect note should state how the choice would change comprehension or action. This short record makes the decision reviewable across languages, editors and document versions.

Pass 18: Run the opposite-link probe

Replace because with although, or so with after, and see how the story changes. The wrong option may sound fluent, revealing why grammar checks alone are inadequate.

Practical record for pass 18: keep an evidence line, a rejected alternative, and a reader-effect note. The evidence line should point to the precise source words, earlier context or approved brief that supports the decision. The rejected alternative should explain which additional premise it would require. The reader-effect note should state how the choice would change comprehension or action. This short record makes the decision reviewable across languages, editors and document versions.

Pass 19: Run the plain-language probe

Ask a neutral reader to identify what the text claims and does not claim. A useful translation should not require readers to solve needless puzzles.

Practical record for pass 19: keep an evidence line, a rejected alternative, and a reader-effect note. The evidence line should point to the precise source words, earlier context or approved brief that supports the decision. The rejected alternative should explain which additional premise it would require. The reader-effect note should state how the choice would change comprehension or action. This short record makes the decision reviewable across languages, editors and document versions.

Pass 20: Run the action probe

If a user is likely to act, test whether they would perform the same action after reading either language. Flag differences that affect safety, money, access or obligations.

Practical record for pass 20: keep an evidence line, a rejected alternative, and a reader-effect note. The evidence line should point to the precise source words, earlier context or approved brief that supports the decision. The rejected alternative should explain which additional premise it would require. The reader-effect note should state how the choice would change comprehension or action. This short record makes the decision reviewable across languages, editors and document versions.

Pass 21: Run the attribution probe

For a claimed reason, identify who supplied the reason and whether the source endorses it. Avoid converting ‘according to the company’ into an unqualified fact.

Practical record for pass 21: keep an evidence line, a rejected alternative, and a reader-effect note. The evidence line should point to the precise source words, earlier context or approved brief that supports the decision. The rejected alternative should explain which additional premise it would require. The reader-effect note should state how the choice would change comprehension or action. This short record makes the decision reviewable across languages, editors and document versions.

Pass 22: Run the corpus probe carefully

Where authentic target-language usage matters, check well-matched texts for natural connective placement and register. Frequency helps choose a form, not prove which source relationship exists.

Practical record for pass 22: keep an evidence line, a rejected alternative, and a reader-effect note. The evidence line should point to the precise source words, earlier context or approved brief that supports the decision. The rejected alternative should explain which additional premise it would require. The reader-effect note should state how the choice would change comprehension or action. This short record makes the decision reviewable across languages, editors and document versions.

Pass 23: Preserve authorial uncertainty

If competing readings survive, ask for clarification or preserve ambiguity. Record unresolved decisions rather than accepting a model’s confident guess.

Practical record for pass 23: keep an evidence line, a rejected alternative, and a reader-effect note. The evidence line should point to the precise source words, earlier context or approved brief that supports the decision. The rejected alternative should explain which additional premise it would require. The reader-effect note should state how the choice would change comprehension or action. This short record makes the decision reviewable across languages, editors and document versions.

Pass 24: Check cross-paragraph continuity

Read two paragraphs before and after the revised sentence. A single however can change the scope of an entire conclusion or cancel a theme established several paragraphs earlier.

Practical record for pass 24: keep an evidence line, a rejected alternative, and a reader-effect note. The evidence line should point to the precise source words, earlier context or approved brief that supports the decision. The rejected alternative should explain which additional premise it would require. The reader-effect note should state how the choice would change comprehension or action. This short record makes the decision reviewable across languages, editors and document versions.

Pass 25: Review source-versus-target relation maps

Make a two-column sheet listing span pairs and relation labels. A new target relation should have an identifiable source basis or a documented adaptation permission.

Practical record for pass 25: keep an evidence line, a rejected alternative, and a reader-effect note. The evidence line should point to the precise source words, earlier context or approved brief that supports the decision. The rejected alternative should explain which additional premise it would require. The reader-effect note should state how the choice would change comprehension or action. This short record makes the decision reviewable across languages, editors and document versions.

Pass 26: Rate severity by consequence

Classify errors in terms of what a reader would infer or do, not by a rote marker list. A small stylistic issue and a false causal claim should not receive equal attention.

Practical record for pass 26: keep an evidence line, a rejected alternative, and a reader-effect note. The evidence line should point to the precise source words, earlier context or approved brief that supports the decision. The rejected alternative should explain which additional premise it would require. The reader-effect note should state how the choice would change comprehension or action. This short record makes the decision reviewable across languages, editors and document versions.

Pass 27: Capture decision provenance

Keep author queries, approved explanations, notes and version references together. This prevents an uncertain connector from becoming an unexamined norm through repeated reuse.

Practical record for pass 27: keep an evidence line, a rejected alternative, and a reader-effect note. The evidence line should point to the precise source words, earlier context or approved brief that supports the decision. The rejected alternative should explain which additional premise it would require. The reader-effect note should state how the choice would change comprehension or action. This short record makes the decision reviewable across languages, editors and document versions.

Pass 28: Read for naturalness once accuracy is secured

After preserving the relation graph, improve rhythm, punctuation, word choice and discourse flow. Naturalness is essential, but it is the final calibration of a sound analysis, not a license to invent meaning.

Practical record for pass 28: keep an evidence line, a rejected alternative, and a reader-effect note. The evidence line should point to the precise source words, earlier context or approved brief that supports the decision. The rejected alternative should explain which additional premise it would require. The reader-effect note should state how the choice would change comprehension or action. This short record makes the decision reviewable across languages, editors and document versions.

4. Original worked diagnostics: 36 documents, six review lenses each

These are controlled training documents, not attested foreign-language quotations. Each case deliberately gives an intended meaning brief so that one can test a target rendering without pretending that juxtaposition alone proves a relation. In production the intended relation must be extracted from the actual source, validated with context, or queried. Every example tests a different consequence of changing a coherence bridge. ‘Unsafe’ means unsafe relative to the supplied meaning brief, not grammatically unallowable in all circumstances.

Case 01: Uncertain laboratory result — Research note

Constructed source meaning: The sensor returned an inconsistent reading. The analyst repeated the measurement.

Intended meaning supplied by the exercise brief: The repeat was a response to uncertainty; no fault was yet diagnosed. Candidate that fails the brief: “The sensor failed, so the analyst had to repeat the test.” Defensible rendering under the brief: “The reading was inconsistent; the analyst repeated the measurement to check it.”

Case 1A. Proposition isolation: what is actually stated

In this research note example, begin with the two source propositions without merging them into a single story: The sensor returned an inconsistent reading. The analyst repeated the measurement. The first analytical task is to preserve each proposition and its evidential status. The passage is not a licence to invent a hidden third event, an extra agent or an unreported decision. A technical anomaly does not prove equipment failure or identify a responsible component. A translator who starts by choosing a smooth connective may overlook that the connective itself carries a claim. Start by marking A and B, their subjects and their temporal or logical scope; then describe the relationship separately in an interpretation note. The relevant relation in the approved training brief is: The repeat was a response to uncertainty; no fault was yet diagnosed.

A useful minimality test is to ask whether each standalone sentence remains true if the proposed relation is false. A second reading could have been part of an ordinary schedule rather than an emergency response. That alternative demonstrates why the words on the page and the link between them require separate auditing. Only the brief for this constructed case resolves the intended link. In an actual translation, identify what language, adjacent document context or author confirmation performs the equivalent role. Do not claim certainty merely because an explanation seems intuitive.

Case 1B. Competing reading and explicitness threshold

Compare the unsafe target, “The sensor failed, so the analyst had to repeat the test.”, with “The reading was inconsistent; the analyst repeated the measurement to check it.”. They are not two equally faithful stylistic options under the stated brief. The former introduces a relationship or implication that needs independent justification; the latter retains the intended boundary. A technical anomaly does not prove equipment failure or identify a responsible component. Test the exact meaning of the linking expression, not merely the dictionary meanings of the nouns and verbs. Ask if the target asserts causation, expectation violation, permission, sequence, evidential proof, or exclusivity. If it does, locate the corresponding source commitment.

Where the real document is genuinely ambiguous, there are three defensible editorial routes. Preserve the ambiguity through natural target phrasing, seek a qualified source-side answer, or use an approved translator note that separates explanation from the translated claim. Do not silently invent a causal explanation to make the paragraph feel complete. The specific competing possibility here is: A second reading could have been part of an ordinary schedule rather than an emergency response. That possibility may or may not be correct, but its continued logical availability shows why an overconfident connective can be dangerous.

Case 1C. Reader-effect and risk diagnostic

A relation error is not just a linguistic blemish; it changes what a reader is licensed to conclude. In this case: A reader could order needless replacement hardware or treat the original value as invalid. Review the target as a person relying on the document rather than as an editor looking for elegant wording. Which action, inference, expectation or accusation becomes more likely after the bad version? What would the intended audience believe if the first source proposition were true but the inferred causal or concessive link were false?

A practical severity review should record the precise sentence, the changed inference, the intended reader, and the possible consequence. If the sentence affects safety, rights, access, technical diagnosis, consent or financial decisions, an unsupported connective deserves more scrutiny than an ordinary style improvement. This scenario’s genre is research note, and the source meaning brief is The repeat was a response to uncertainty; no fault was yet diagnosed. That is the benchmark for both the accuracy review and the final plain-language edit.

Case 1D. Evidence and author-query gate

An efficient reviewer does not write ‘unclear’ and stop. Ask a question that can be answered with evidence: Did the analyst repeat for verification, because of a verified malfunction, or for another reason? Include the source passage and the two candidate target readings. An author can then confirm the relation, provide missing context or explicitly leave the ambiguity open. The record should distinguish an author-confirmed meaning from a translator’s plausible interpretation and from a machine-generated suggestion.

After the answer, update the relation map, not only the visible sentence. If similar segments occur elsewhere, check them for consistency without mechanically propagating the same connective. Context may legitimately make the same pair of vocabulary items express a different relationship in a later document. For this exercise, the brief already supplies the acceptable decision: The repeat was a response to uncertainty; no fault was yet diagnosed. The next step is to verify that the chosen target preserves it and creates no new attribution.

Case 1E. Adversarial paraphrase and back-reading

Take the faithful candidate — “The reading was inconsistent; the analyst repeated the measurement to check it.” — and paraphrase it into a simple reader belief. Now take the unsafe candidate — “The sensor failed, so the analyst had to repeat the test.” — and do the same. Write down the extra proposition, presupposition or causal claim that appears in the second version. A technical anomaly does not prove equipment failure or identify a responsible component. This comparison is stronger than asking whether both sentences sound natural. A fluent sentence can be misleading, and a back-translation can repeat the same mistake if the reviewer already accepts its assumptions.

For an adversarial check, imagine a reader who accepts the two reported events but rejects the suggested link. Would that reader still be entitled to believe the source? Consider the possibility that a second reading could have been part of an ordinary schedule rather than an emergency response. If yes, the stronger reading needs explicit textual or authorial support. The test does not require mathematical proof. It disciplines translation by making the asserted relationship visible and by preventing an intuition from hiding inside a connective.

Case 1F. Transfer drill and acceptance criterion

Transfer the same reasoning to another language, medium or genre without carrying over English word order. Keep the propositional facts and the stated relation fixed, but choose a natural construction suitable for the target audience. The final acceptable solution does not have to use the English connector in “The reading was inconsistent; the analyst repeated the measurement to check it.” literally. It does have to preserve the repeat was a response to uncertainty; no fault was yet diagnosed. It must also avoid the new belief introduced by “The sensor failed, so the analyst had to repeat the test.”.

The review is passed when a target reader can identify the two facts, draw only the source-authorized inference, and answer the practical question raised by this research note text without adopting an unsupported explanation. The specific failure warning is: A reader could order needless replacement hardware or treat the original value as invalid. Keep the approval record with this sentence’s context, the reviewer’s decision and any author response. If a future revision changes the source relation, rerun this check rather than treating the existing translation as automatically safe.

Case 02: Reopened community room — Public notice

Constructed source meaning: The large room reopened. The small studio remained closed.

Intended meaning supplied by the exercise brief: The first reopening does not extend to the second space; the second sentence is a scope qualification. Candidate that fails the brief: “The entire facility reopened because repairs were complete.” Defensible rendering under the brief: “The large room reopened, but the small studio remained closed.”

Case 2A. Proposition isolation: what is actually stated

In this public notice example, begin with the two source propositions without merging them into a single story: The large room reopened. The small studio remained closed. The first analytical task is to preserve each proposition and its evidential status. The passage is not a licence to invent a hidden third event, an extra agent or an unreported decision. The target must preserve the difference between partial and full access. A translator who starts by choosing a smooth connective may overlook that the connective itself carries a claim. Start by marking A and B, their subjects and their temporal or logical scope; then describe the relationship separately in an interpretation note. The relevant relation in the approved training brief is: The first reopening does not extend to the second space; the second sentence is a scope qualification.

A useful minimality test is to ask whether each standalone sentence remains true if the proposed relation is false. The two rooms could have separate reopening schedules. That alternative demonstrates why the words on the page and the link between them require separate auditing. Only the brief for this constructed case resolves the intended link. In an actual translation, identify what language, adjacent document context or author confirmation performs the equivalent role. Do not claim certainty merely because an explanation seems intuitive.

Case 2B. Competing reading and explicitness threshold

Compare the unsafe target, “The entire facility reopened because repairs were complete.”, with “The large room reopened, but the small studio remained closed.”. They are not two equally faithful stylistic options under the stated brief. The former introduces a relationship or implication that needs independent justification; the latter retains the intended boundary. The target must preserve the difference between partial and full access. Test the exact meaning of the linking expression, not merely the dictionary meanings of the nouns and verbs. Ask if the target asserts causation, expectation violation, permission, sequence, evidential proof, or exclusivity. If it does, locate the corresponding source commitment.

Where the real document is genuinely ambiguous, there are three defensible editorial routes. Preserve the ambiguity through natural target phrasing, seek a qualified source-side answer, or use an approved translator note that separates explanation from the translated claim. Do not silently invent a causal explanation to make the paragraph feel complete. The specific competing possibility here is: The two rooms could have separate reopening schedules. That possibility may or may not be correct, but its continued logical availability shows why an overconfident connective can be dangerous.

Case 2C. Reader-effect and risk diagnostic

A relation error is not just a linguistic blemish; it changes what a reader is licensed to conclude. In this case: Visitors could arrive expecting facilities that are still unavailable. Review the target as a person relying on the document rather than as an editor looking for elegant wording. Which action, inference, expectation or accusation becomes more likely after the bad version? What would the intended audience believe if the first source proposition were true but the inferred causal or concessive link were false?

A practical severity review should record the precise sentence, the changed inference, the intended reader, and the possible consequence. If the sentence affects safety, rights, access, technical diagnosis, consent or financial decisions, an unsupported connective deserves more scrutiny than an ordinary style improvement. This scenario’s genre is public notice, and the source meaning brief is The first reopening does not extend to the second space; the second sentence is a scope qualification. That is the benchmark for both the accuracy review and the final plain-language edit.

Case 2D. Evidence and author-query gate

An efficient reviewer does not write ‘unclear’ and stop. Ask a question that can be answered with evidence: Is the second sentence an exception to a general opening notice or an independent status update? Include the source passage and the two candidate target readings. An author can then confirm the relation, provide missing context or explicitly leave the ambiguity open. The record should distinguish an author-confirmed meaning from a translator’s plausible interpretation and from a machine-generated suggestion.

After the answer, update the relation map, not only the visible sentence. If similar segments occur elsewhere, check them for consistency without mechanically propagating the same connective. Context may legitimately make the same pair of vocabulary items express a different relationship in a later document. For this exercise, the brief already supplies the acceptable decision: The first reopening does not extend to the second space; the second sentence is a scope qualification. The next step is to verify that the chosen target preserves it and creates no new attribution.

Case 2E. Adversarial paraphrase and back-reading

Take the faithful candidate — “The large room reopened, but the small studio remained closed.” — and paraphrase it into a simple reader belief. Now take the unsafe candidate — “The entire facility reopened because repairs were complete.” — and do the same. Write down the extra proposition, presupposition or causal claim that appears in the second version. The target must preserve the difference between partial and full access. This comparison is stronger than asking whether both sentences sound natural. A fluent sentence can be misleading, and a back-translation can repeat the same mistake if the reviewer already accepts its assumptions.

For an adversarial check, imagine a reader who accepts the two reported events but rejects the suggested link. Would that reader still be entitled to believe the source? Consider the possibility that the two rooms could have separate reopening schedules. If yes, the stronger reading needs explicit textual or authorial support. The test does not require mathematical proof. It disciplines translation by making the asserted relationship visible and by preventing an intuition from hiding inside a connective.

Case 2F. Transfer drill and acceptance criterion

Transfer the same reasoning to another language, medium or genre without carrying over English word order. Keep the propositional facts and the stated relation fixed, but choose a natural construction suitable for the target audience. The final acceptable solution does not have to use the English connector in “The large room reopened, but the small studio remained closed.” literally. It does have to preserve the first reopening does not extend to the second space; the second sentence is a scope qualification. It must also avoid the new belief introduced by “The entire facility reopened because repairs were complete.”.

The review is passed when a target reader can identify the two facts, draw only the source-authorized inference, and answer the practical question raised by this public notice text without adopting an unsupported explanation. The specific failure warning is: Visitors could arrive expecting facilities that are still unavailable. Keep the approval record with this sentence’s context, the reviewer’s decision and any author response. If a future revision changes the source relation, rerun this check rather than treating the existing translation as automatically safe.

Case 03: Maintenance and service delay — Status page

Constructed source meaning: Maintenance finished at dawn. Service resumed in the afternoon.

Intended meaning supplied by the exercise brief: Sequence is known; the interval is not by itself proof of why service was delayed. Candidate that fails the brief: “Service resumed late because maintenance ran until afternoon.” Defensible rendering under the brief: “Maintenance finished at dawn; service resumed in the afternoon.”

Case 3A. Proposition isolation: what is actually stated

In this status page example, begin with the two source propositions without merging them into a single story: Maintenance finished at dawn. Service resumed in the afternoon. The first analytical task is to preserve each proposition and its evidential status. The passage is not a licence to invent a hidden third event, an extra agent or an unreported decision. The bad version changes the maintenance end time and asserts a cause that is not established. A translator who starts by choosing a smooth connective may overlook that the connective itself carries a claim. Start by marking A and B, their subjects and their temporal or logical scope; then describe the relationship separately in an interpretation note. The relevant relation in the approved training brief is: Sequence is known; the interval is not by itself proof of why service was delayed.

A useful minimality test is to ask whether each standalone sentence remains true if the proposed relation is false. Completion of maintenance could precede an unrelated service issue. That alternative demonstrates why the words on the page and the link between them require separate auditing. Only the brief for this constructed case resolves the intended link. In an actual translation, identify what language, adjacent document context or author confirmation performs the equivalent role. Do not claim certainty merely because an explanation seems intuitive.

Case 3B. Competing reading and explicitness threshold

Compare the unsafe target, “Service resumed late because maintenance ran until afternoon.”, with “Maintenance finished at dawn; service resumed in the afternoon.”. They are not two equally faithful stylistic options under the stated brief. The former introduces a relationship or implication that needs independent justification; the latter retains the intended boundary. The bad version changes the maintenance end time and asserts a cause that is not established. Test the exact meaning of the linking expression, not merely the dictionary meanings of the nouns and verbs. Ask if the target asserts causation, expectation violation, permission, sequence, evidential proof, or exclusivity. If it does, locate the corresponding source commitment.

Where the real document is genuinely ambiguous, there are three defensible editorial routes. Preserve the ambiguity through natural target phrasing, seek a qualified source-side answer, or use an approved translator note that separates explanation from the translated claim. Do not silently invent a causal explanation to make the paragraph feel complete. The specific competing possibility here is: Completion of maintenance could precede an unrelated service issue. That possibility may or may not be correct, but its continued logical availability shows why an overconfident connective can be dangerous.

Case 3C. Reader-effect and risk diagnostic

A relation error is not just a linguistic blemish; it changes what a reader is licensed to conclude. In this case: Users may infer an operator missed a maintenance deadline. Review the target as a person relying on the document rather than as an editor looking for elegant wording. Which action, inference, expectation or accusation becomes more likely after the bad version? What would the intended audience believe if the first source proposition were true but the inferred causal or concessive link were false?

A practical severity review should record the precise sentence, the changed inference, the intended reader, and the possible consequence. If the sentence affects safety, rights, access, technical diagnosis, consent or financial decisions, an unsupported connective deserves more scrutiny than an ordinary style improvement. This scenario’s genre is status page, and the source meaning brief is Sequence is known; the interval is not by itself proof of why service was delayed. That is the benchmark for both the accuracy review and the final plain-language edit.

Case 3D. Evidence and author-query gate

An efficient reviewer does not write ‘unclear’ and stop. Ask a question that can be answered with evidence: Was testing, authorization or another event responsible for the interval? Include the source passage and the two candidate target readings. An author can then confirm the relation, provide missing context or explicitly leave the ambiguity open. The record should distinguish an author-confirmed meaning from a translator’s plausible interpretation and from a machine-generated suggestion.

After the answer, update the relation map, not only the visible sentence. If similar segments occur elsewhere, check them for consistency without mechanically propagating the same connective. Context may legitimately make the same pair of vocabulary items express a different relationship in a later document. For this exercise, the brief already supplies the acceptable decision: Sequence is known; the interval is not by itself proof of why service was delayed. The next step is to verify that the chosen target preserves it and creates no new attribution.

Case 3E. Adversarial paraphrase and back-reading

Take the faithful candidate — “Maintenance finished at dawn; service resumed in the afternoon.” — and paraphrase it into a simple reader belief. Now take the unsafe candidate — “Service resumed late because maintenance ran until afternoon.” — and do the same. Write down the extra proposition, presupposition or causal claim that appears in the second version. The bad version changes the maintenance end time and asserts a cause that is not established. This comparison is stronger than asking whether both sentences sound natural. A fluent sentence can be misleading, and a back-translation can repeat the same mistake if the reviewer already accepts its assumptions.

For an adversarial check, imagine a reader who accepts the two reported events but rejects the suggested link. Would that reader still be entitled to believe the source? Consider the possibility that completion of maintenance could precede an unrelated service issue. If yes, the stronger reading needs explicit textual or authorial support. The test does not require mathematical proof. It disciplines translation by making the asserted relationship visible and by preventing an intuition from hiding inside a connective.

Case 3F. Transfer drill and acceptance criterion

Transfer the same reasoning to another language, medium or genre without carrying over English word order. Keep the propositional facts and the stated relation fixed, but choose a natural construction suitable for the target audience. The final acceptable solution does not have to use the English connector in “Maintenance finished at dawn; service resumed in the afternoon.” literally. It does have to preserve sequence is known; the interval is not by itself proof of why service was delayed. It must also avoid the new belief introduced by “Service resumed late because maintenance ran until afternoon.”.

The review is passed when a target reader can identify the two facts, draw only the source-authorized inference, and answer the practical question raised by this status page text without adopting an unsupported explanation. The specific failure warning is: Users may infer an operator missed a maintenance deadline. Keep the approval record with this sentence’s context, the reviewer’s decision and any author response. If a future revision changes the source relation, rerun this check rather than treating the existing translation as automatically safe.

Case 04: Small dataset and restrained conclusion — Research abstract

Constructed source meaning: The pilot sample was small. The team described the findings as preliminary.

Intended meaning supplied by the exercise brief: The author brief attributes cautious wording to evidence limits without claiming every conclusion was wrong. Candidate that fails the brief: “The tiny sample proved the research useless.” Defensible rendering under the brief: “Because the pilot sample was small, the team described the findings as preliminary.”

Case 4A. Proposition isolation: what is actually stated

In this research abstract example, begin with the two source propositions without merging them into a single story: The pilot sample was small. The team described the findings as preliminary. The first analytical task is to preserve each proposition and its evidential status. The passage is not a licence to invent a hidden third event, an extra agent or an unreported decision. A limitation that warrants caution does not entail a null result or scientific worthlessness. A translator who starts by choosing a smooth connective may overlook that the connective itself carries a claim. Start by marking A and B, their subjects and their temporal or logical scope; then describe the relationship separately in an interpretation note. The relevant relation in the approved training brief is: The author brief attributes cautious wording to evidence limits without claiming every conclusion was wrong.

A useful minimality test is to ask whether each standalone sentence remains true if the proposed relation is false. Other limitations might also justify a preliminary label. That alternative demonstrates why the words on the page and the link between them require separate auditing. Only the brief for this constructed case resolves the intended link. In an actual translation, identify what language, adjacent document context or author confirmation performs the equivalent role. Do not claim certainty merely because an explanation seems intuitive.

Case 4B. Competing reading and explicitness threshold

Compare the unsafe target, “The tiny sample proved the research useless.”, with “Because the pilot sample was small, the team described the findings as preliminary.”. They are not two equally faithful stylistic options under the stated brief. The former introduces a relationship or implication that needs independent justification; the latter retains the intended boundary. A limitation that warrants caution does not entail a null result or scientific worthlessness. Test the exact meaning of the linking expression, not merely the dictionary meanings of the nouns and verbs. Ask if the target asserts causation, expectation violation, permission, sequence, evidential proof, or exclusivity. If it does, locate the corresponding source commitment.

Where the real document is genuinely ambiguous, there are three defensible editorial routes. Preserve the ambiguity through natural target phrasing, seek a qualified source-side answer, or use an approved translator note that separates explanation from the translated claim. Do not silently invent a causal explanation to make the paragraph feel complete. The specific competing possibility here is: Other limitations might also justify a preliminary label. That possibility may or may not be correct, but its continued logical availability shows why an overconfident connective can be dangerous.

Case 4C. Reader-effect and risk diagnostic

A relation error is not just a linguistic blemish; it changes what a reader is licensed to conclude. In this case: Readers could reject useful preliminary evidence or treat a pilot as a definitive trial. Review the target as a person relying on the document rather than as an editor looking for elegant wording. Which action, inference, expectation or accusation becomes more likely after the bad version? What would the intended audience believe if the first source proposition were true but the inferred causal or concessive link were false?

A practical severity review should record the precise sentence, the changed inference, the intended reader, and the possible consequence. If the sentence affects safety, rights, access, technical diagnosis, consent or financial decisions, an unsupported connective deserves more scrutiny than an ordinary style improvement. This scenario’s genre is research abstract, and the source meaning brief is The author brief attributes cautious wording to evidence limits without claiming every conclusion was wrong. That is the benchmark for both the accuracy review and the final plain-language edit.

Case 4D. Evidence and author-query gate

An efficient reviewer does not write ‘unclear’ and stop. Ask a question that can be answered with evidence: Is sample size the stated reason for the preliminary designation? Include the source passage and the two candidate target readings. An author can then confirm the relation, provide missing context or explicitly leave the ambiguity open. The record should distinguish an author-confirmed meaning from a translator’s plausible interpretation and from a machine-generated suggestion.

After the answer, update the relation map, not only the visible sentence. If similar segments occur elsewhere, check them for consistency without mechanically propagating the same connective. Context may legitimately make the same pair of vocabulary items express a different relationship in a later document. For this exercise, the brief already supplies the acceptable decision: The author brief attributes cautious wording to evidence limits without claiming every conclusion was wrong. The next step is to verify that the chosen target preserves it and creates no new attribution.

Case 4E. Adversarial paraphrase and back-reading

Take the faithful candidate — “Because the pilot sample was small, the team described the findings as preliminary.” — and paraphrase it into a simple reader belief. Now take the unsafe candidate — “The tiny sample proved the research useless.” — and do the same. Write down the extra proposition, presupposition or causal claim that appears in the second version. A limitation that warrants caution does not entail a null result or scientific worthlessness. This comparison is stronger than asking whether both sentences sound natural. A fluent sentence can be misleading, and a back-translation can repeat the same mistake if the reviewer already accepts its assumptions.

For an adversarial check, imagine a reader who accepts the two reported events but rejects the suggested link. Would that reader still be entitled to believe the source? Consider the possibility that other limitations might also justify a preliminary label. If yes, the stronger reading needs explicit textual or authorial support. The test does not require mathematical proof. It disciplines translation by making the asserted relationship visible and by preventing an intuition from hiding inside a connective.

Case 4F. Transfer drill and acceptance criterion

Transfer the same reasoning to another language, medium or genre without carrying over English word order. Keep the propositional facts and the stated relation fixed, but choose a natural construction suitable for the target audience. The final acceptable solution does not have to use the English connector in “Because the pilot sample was small, the team described the findings as preliminary.” literally. It does have to preserve the author brief attributes cautious wording to evidence limits without claiming every conclusion was wrong. It must also avoid the new belief introduced by “The tiny sample proved the research useless.”.

The review is passed when a target reader can identify the two facts, draw only the source-authorized inference, and answer the practical question raised by this research abstract text without adopting an unsupported explanation. The specific failure warning is: Readers could reject useful preliminary evidence or treat a pilot as a definitive trial. Keep the approval record with this sentence’s context, the reviewer’s decision and any author response. If a future revision changes the source relation, rerun this check rather than treating the existing translation as automatically safe.

Case 05: Training and selection — Biography

Constructed source meaning: The candidate trained daily. The panel selected someone else.

Intended meaning supplied by the exercise brief: An expectation-defeating concession is intended, not a claim that training causes rejection. Candidate that fails the brief: “The panel rejected the candidate because of daily training.” Defensible rendering under the brief: “Although the candidate trained daily, the panel selected someone else.”

Case 5A. Proposition isolation: what is actually stated

In this biography example, begin with the two source propositions without merging them into a single story: The candidate trained daily. The panel selected someone else. The first analytical task is to preserve each proposition and its evidential status. The passage is not a licence to invent a hidden third event, an extra agent or an unreported decision. The wrong causal bridge reverses the evaluative force of the contrast. A translator who starts by choosing a smooth connective may overlook that the connective itself carries a claim. Start by marking A and B, their subjects and their temporal or logical scope; then describe the relationship separately in an interpretation note. The relevant relation in the approved training brief is: An expectation-defeating concession is intended, not a claim that training causes rejection.

A useful minimality test is to ask whether each standalone sentence remains true if the proposed relation is false. The selection could have turned on eligibility unknown to the writer. That alternative demonstrates why the words on the page and the link between them require separate auditing. Only the brief for this constructed case resolves the intended link. In an actual translation, identify what language, adjacent document context or author confirmation performs the equivalent role. Do not claim certainty merely because an explanation seems intuitive.

Case 5B. Competing reading and explicitness threshold

Compare the unsafe target, “The panel rejected the candidate because of daily training.”, with “Although the candidate trained daily, the panel selected someone else.”. They are not two equally faithful stylistic options under the stated brief. The former introduces a relationship or implication that needs independent justification; the latter retains the intended boundary. The wrong causal bridge reverses the evaluative force of the contrast. Test the exact meaning of the linking expression, not merely the dictionary meanings of the nouns and verbs. Ask if the target asserts causation, expectation violation, permission, sequence, evidential proof, or exclusivity. If it does, locate the corresponding source commitment.

Where the real document is genuinely ambiguous, there are three defensible editorial routes. Preserve the ambiguity through natural target phrasing, seek a qualified source-side answer, or use an approved translator note that separates explanation from the translated claim. Do not silently invent a causal explanation to make the paragraph feel complete. The specific competing possibility here is: The selection could have turned on eligibility unknown to the writer. That possibility may or may not be correct, but its continued logical availability shows why an overconfident connective can be dangerous.

Case 5C. Reader-effect and risk diagnostic

A relation error is not just a linguistic blemish; it changes what a reader is licensed to conclude. In this case: Readers may believe diligence was a disadvantage rather than simply insufficient. Review the target as a person relying on the document rather than as an editor looking for elegant wording. Which action, inference, expectation or accusation becomes more likely after the bad version? What would the intended audience believe if the first source proposition were true but the inferred causal or concessive link were false?

A practical severity review should record the precise sentence, the changed inference, the intended reader, and the possible consequence. If the sentence affects safety, rights, access, technical diagnosis, consent or financial decisions, an unsupported connective deserves more scrutiny than an ordinary style improvement. This scenario’s genre is biography, and the source meaning brief is An expectation-defeating concession is intended, not a claim that training causes rejection. That is the benchmark for both the accuracy review and the final plain-language edit.

Case 5D. Evidence and author-query gate

An efficient reviewer does not write ‘unclear’ and stop. Ask a question that can be answered with evidence: Does the author intend surprise, or just chronological biography? Include the source passage and the two candidate target readings. An author can then confirm the relation, provide missing context or explicitly leave the ambiguity open. The record should distinguish an author-confirmed meaning from a translator’s plausible interpretation and from a machine-generated suggestion.

After the answer, update the relation map, not only the visible sentence. If similar segments occur elsewhere, check them for consistency without mechanically propagating the same connective. Context may legitimately make the same pair of vocabulary items express a different relationship in a later document. For this exercise, the brief already supplies the acceptable decision: An expectation-defeating concession is intended, not a claim that training causes rejection. The next step is to verify that the chosen target preserves it and creates no new attribution.

Case 5E. Adversarial paraphrase and back-reading

Take the faithful candidate — “Although the candidate trained daily, the panel selected someone else.” — and paraphrase it into a simple reader belief. Now take the unsafe candidate — “The panel rejected the candidate because of daily training.” — and do the same. Write down the extra proposition, presupposition or causal claim that appears in the second version. The wrong causal bridge reverses the evaluative force of the contrast. This comparison is stronger than asking whether both sentences sound natural. A fluent sentence can be misleading, and a back-translation can repeat the same mistake if the reviewer already accepts its assumptions.

For an adversarial check, imagine a reader who accepts the two reported events but rejects the suggested link. Would that reader still be entitled to believe the source? Consider the possibility that the selection could have turned on eligibility unknown to the writer. If yes, the stronger reading needs explicit textual or authorial support. The test does not require mathematical proof. It disciplines translation by making the asserted relationship visible and by preventing an intuition from hiding inside a connective.

Case 5F. Transfer drill and acceptance criterion

Transfer the same reasoning to another language, medium or genre without carrying over English word order. Keep the propositional facts and the stated relation fixed, but choose a natural construction suitable for the target audience. The final acceptable solution does not have to use the English connector in “Although the candidate trained daily, the panel selected someone else.” literally. It does have to preserve an expectation-defeating concession is intended, not a claim that training causes rejection. It must also avoid the new belief introduced by “The panel rejected the candidate because of daily training.”.

The review is passed when a target reader can identify the two facts, draw only the source-authorized inference, and answer the practical question raised by this biography text without adopting an unsupported explanation. The specific failure warning is: Readers may believe diligence was a disadvantage rather than simply insufficient. Keep the approval record with this sentence’s context, the reviewer’s decision and any author response. If a future revision changes the source relation, rerun this check rather than treating the existing translation as automatically safe.

Case 06: Extra rainfall and reduced yield — Field report

Constructed source meaning: Rainfall increased. The harvest yield fell.

Intended meaning supplied by the exercise brief: The source contrasts observations without identifying a causal pathway. Candidate that fails the brief: “Yield fell because rainfall increased.” Defensible rendering under the brief: “Rainfall increased, while harvest yield fell.”

Case 6A. Proposition isolation: what is actually stated

In this field report example, begin with the two source propositions without merging them into a single story: Rainfall increased. The harvest yield fell. The first analytical task is to preserve each proposition and its evidential status. The passage is not a licence to invent a hidden third event, an extra agent or an unreported decision. The source cannot support a single-factor causal explanation from two observations. A translator who starts by choosing a smooth connective may overlook that the connective itself carries a claim. Start by marking A and B, their subjects and their temporal or logical scope; then describe the relationship separately in an interpretation note. The relevant relation in the approved training brief is: The source contrasts observations without identifying a causal pathway.

A useful minimality test is to ask whether each standalone sentence remains true if the proposed relation is false. Pests or temperature could explain the yield independently. That alternative demonstrates why the words on the page and the link between them require separate auditing. Only the brief for this constructed case resolves the intended link. In an actual translation, identify what language, adjacent document context or author confirmation performs the equivalent role. Do not claim certainty merely because an explanation seems intuitive.

Case 6B. Competing reading and explicitness threshold

Compare the unsafe target, “Yield fell because rainfall increased.”, with “Rainfall increased, while harvest yield fell.”. They are not two equally faithful stylistic options under the stated brief. The former introduces a relationship or implication that needs independent justification; the latter retains the intended boundary. The source cannot support a single-factor causal explanation from two observations. Test the exact meaning of the linking expression, not merely the dictionary meanings of the nouns and verbs. Ask if the target asserts causation, expectation violation, permission, sequence, evidential proof, or exclusivity. If it does, locate the corresponding source commitment.

Where the real document is genuinely ambiguous, there are three defensible editorial routes. Preserve the ambiguity through natural target phrasing, seek a qualified source-side answer, or use an approved translator note that separates explanation from the translated claim. Do not silently invent a causal explanation to make the paragraph feel complete. The specific competing possibility here is: Pests or temperature could explain the yield independently. That possibility may or may not be correct, but its continued logical availability shows why an overconfident connective can be dangerous.

Case 6C. Reader-effect and risk diagnostic

A relation error is not just a linguistic blemish; it changes what a reader is licensed to conclude. In this case: Agricultural readers may draw an unsupported management conclusion. Review the target as a person relying on the document rather than as an editor looking for elegant wording. Which action, inference, expectation or accusation becomes more likely after the bad version? What would the intended audience believe if the first source proposition were true but the inferred causal or concessive link were false?

A practical severity review should record the precise sentence, the changed inference, the intended reader, and the possible consequence. If the sentence affects safety, rights, access, technical diagnosis, consent or financial decisions, an unsupported connective deserves more scrutiny than an ordinary style improvement. This scenario’s genre is field report, and the source meaning brief is The source contrasts observations without identifying a causal pathway. That is the benchmark for both the accuracy review and the final plain-language edit.

Case 6D. Evidence and author-query gate

An efficient reviewer does not write ‘unclear’ and stop. Ask a question that can be answered with evidence: Is there separate evidence of waterlogging or another mechanism? Include the source passage and the two candidate target readings. An author can then confirm the relation, provide missing context or explicitly leave the ambiguity open. The record should distinguish an author-confirmed meaning from a translator’s plausible interpretation and from a machine-generated suggestion.

After the answer, update the relation map, not only the visible sentence. If similar segments occur elsewhere, check them for consistency without mechanically propagating the same connective. Context may legitimately make the same pair of vocabulary items express a different relationship in a later document. For this exercise, the brief already supplies the acceptable decision: The source contrasts observations without identifying a causal pathway. The next step is to verify that the chosen target preserves it and creates no new attribution.

Case 6E. Adversarial paraphrase and back-reading

Take the faithful candidate — “Rainfall increased, while harvest yield fell.” — and paraphrase it into a simple reader belief. Now take the unsafe candidate — “Yield fell because rainfall increased.” — and do the same. Write down the extra proposition, presupposition or causal claim that appears in the second version. The source cannot support a single-factor causal explanation from two observations. This comparison is stronger than asking whether both sentences sound natural. A fluent sentence can be misleading, and a back-translation can repeat the same mistake if the reviewer already accepts its assumptions.

For an adversarial check, imagine a reader who accepts the two reported events but rejects the suggested link. Would that reader still be entitled to believe the source? Consider the possibility that pests or temperature could explain the yield independently. If yes, the stronger reading needs explicit textual or authorial support. The test does not require mathematical proof. It disciplines translation by making the asserted relationship visible and by preventing an intuition from hiding inside a connective.

Case 6F. Transfer drill and acceptance criterion

Transfer the same reasoning to another language, medium or genre without carrying over English word order. Keep the propositional facts and the stated relation fixed, but choose a natural construction suitable for the target audience. The final acceptable solution does not have to use the English connector in “Rainfall increased, while harvest yield fell.” literally. It does have to preserve the source contrasts observations without identifying a causal pathway. It must also avoid the new belief introduced by “Yield fell because rainfall increased.”.

The review is passed when a target reader can identify the two facts, draw only the source-authorized inference, and answer the practical question raised by this field report text without adopting an unsupported explanation. The specific failure warning is: Agricultural readers may draw an unsupported management conclusion. Keep the approval record with this sentence’s context, the reviewer’s decision and any author response. If a future revision changes the source relation, rerun this check rather than treating the existing translation as automatically safe.

Case 07: Offline feature availability — Product description

Constructed source meaning: The map remained visible. The network connection was lost.

Intended meaning supplied by the exercise brief: The text expresses a robust functionality surviving a normally relevant limitation. Candidate that fails the brief: “The map was visible only because the connection failed.” Defensible rendering under the brief: “The connection was lost, yet the map remained visible.”

Case 7A. Proposition isolation: what is actually stated

In this product description example, begin with the two source propositions without merging them into a single story: The map remained visible. The network connection was lost. The first analytical task is to preserve each proposition and its evidential status. The passage is not a licence to invent a hidden third event, an extra agent or an unreported decision. The unsafe direction turns resilience into a dependency on failure. A translator who starts by choosing a smooth connective may overlook that the connective itself carries a claim. Start by marking A and B, their subjects and their temporal or logical scope; then describe the relationship separately in an interpretation note. The relevant relation in the approved training brief is: The text expresses a robust functionality surviving a normally relevant limitation.

A useful minimality test is to ask whether each standalone sentence remains true if the proposed relation is false. Some versions may not make every layer available offline. That alternative demonstrates why the words on the page and the link between them require separate auditing. Only the brief for this constructed case resolves the intended link. In an actual translation, identify what language, adjacent document context or author confirmation performs the equivalent role. Do not claim certainty merely because an explanation seems intuitive.

Case 7B. Competing reading and explicitness threshold

Compare the unsafe target, “The map was visible only because the connection failed.”, with “The connection was lost, yet the map remained visible.”. They are not two equally faithful stylistic options under the stated brief. The former introduces a relationship or implication that needs independent justification; the latter retains the intended boundary. The unsafe direction turns resilience into a dependency on failure. Test the exact meaning of the linking expression, not merely the dictionary meanings of the nouns and verbs. Ask if the target asserts causation, expectation violation, permission, sequence, evidential proof, or exclusivity. If it does, locate the corresponding source commitment.

Where the real document is genuinely ambiguous, there are three defensible editorial routes. Preserve the ambiguity through natural target phrasing, seek a qualified source-side answer, or use an approved translator note that separates explanation from the translated claim. Do not silently invent a causal explanation to make the paragraph feel complete. The specific competing possibility here is: Some versions may not make every layer available offline. That possibility may or may not be correct, but its continued logical availability shows why an overconfident connective can be dangerous.

Case 7C. Reader-effect and risk diagnostic

A relation error is not just a linguistic blemish; it changes what a reader is licensed to conclude. In this case: A reader may wrongly switch off connectivity to use the map. Review the target as a person relying on the document rather than as an editor looking for elegant wording. Which action, inference, expectation or accusation becomes more likely after the bad version? What would the intended audience believe if the first source proposition were true but the inferred causal or concessive link were false?

A practical severity review should record the precise sentence, the changed inference, the intended reader, and the possible consequence. If the sentence affects safety, rights, access, technical diagnosis, consent or financial decisions, an unsupported connective deserves more scrutiny than an ordinary style improvement. This scenario’s genre is product description, and the source meaning brief is The text expresses a robust functionality surviving a normally relevant limitation. That is the benchmark for both the accuracy review and the final plain-language edit.

Case 7D. Evidence and author-query gate

An efficient reviewer does not write ‘unclear’ and stop. Ask a question that can be answered with evidence: Does the map require cached data or a separate offline package? Include the source passage and the two candidate target readings. An author can then confirm the relation, provide missing context or explicitly leave the ambiguity open. The record should distinguish an author-confirmed meaning from a translator’s plausible interpretation and from a machine-generated suggestion.

After the answer, update the relation map, not only the visible sentence. If similar segments occur elsewhere, check them for consistency without mechanically propagating the same connective. Context may legitimately make the same pair of vocabulary items express a different relationship in a later document. For this exercise, the brief already supplies the acceptable decision: The text expresses a robust functionality surviving a normally relevant limitation. The next step is to verify that the chosen target preserves it and creates no new attribution.

Case 7E. Adversarial paraphrase and back-reading

Take the faithful candidate — “The connection was lost, yet the map remained visible.” — and paraphrase it into a simple reader belief. Now take the unsafe candidate — “The map was visible only because the connection failed.” — and do the same. Write down the extra proposition, presupposition or causal claim that appears in the second version. The unsafe direction turns resilience into a dependency on failure. This comparison is stronger than asking whether both sentences sound natural. A fluent sentence can be misleading, and a back-translation can repeat the same mistake if the reviewer already accepts its assumptions.

For an adversarial check, imagine a reader who accepts the two reported events but rejects the suggested link. Would that reader still be entitled to believe the source? Consider the possibility that some versions may not make every layer available offline. If yes, the stronger reading needs explicit textual or authorial support. The test does not require mathematical proof. It disciplines translation by making the asserted relationship visible and by preventing an intuition from hiding inside a connective.

Case 7F. Transfer drill and acceptance criterion

Transfer the same reasoning to another language, medium or genre without carrying over English word order. Keep the propositional facts and the stated relation fixed, but choose a natural construction suitable for the target audience. The final acceptable solution does not have to use the English connector in “The connection was lost, yet the map remained visible.” literally. It does have to preserve the text expresses a robust functionality surviving a normally relevant limitation. It must also avoid the new belief introduced by “The map was visible only because the connection failed.”.

The review is passed when a target reader can identify the two facts, draw only the source-authorized inference, and answer the practical question raised by this product description text without adopting an unsupported explanation. The specific failure warning is: A reader may wrongly switch off connectivity to use the map. Keep the approval record with this sentence’s context, the reviewer’s decision and any author response. If a future revision changes the source relation, rerun this check rather than treating the existing translation as automatically safe.

Case 08: Meeting without a quorum — Meeting record

Constructed source meaning: Fewer members attended than required. The meeting could not vote.

Intended meaning supplied by the exercise brief: A rule-based dependency is confirmed; discussion may still occur. Candidate that fails the brief: “The meeting was cancelled and no business took place.” Defensible rendering under the brief: “Without the required quorum, the meeting could not vote.”

Case 8A. Proposition isolation: what is actually stated

In this meeting record example, begin with the two source propositions without merging them into a single story: Fewer members attended than required. The meeting could not vote. The first analytical task is to preserve each proposition and its evidential status. The passage is not a licence to invent a hidden third event, an extra agent or an unreported decision. Inability to vote is not identical to cancellation of the gathering. A translator who starts by choosing a smooth connective may overlook that the connective itself carries a claim. Start by marking A and B, their subjects and their temporal or logical scope; then describe the relationship separately in an interpretation note. The relevant relation in the approved training brief is: A rule-based dependency is confirmed; discussion may still occur.

A useful minimality test is to ask whether each standalone sentence remains true if the proposed relation is false. Some organizations permit discussion even when binding decisions are impossible. That alternative demonstrates why the words on the page and the link between them require separate auditing. Only the brief for this constructed case resolves the intended link. In an actual translation, identify what language, adjacent document context or author confirmation performs the equivalent role. Do not claim certainty merely because an explanation seems intuitive.

Case 8B. Competing reading and explicitness threshold

Compare the unsafe target, “The meeting was cancelled and no business took place.”, with “Without the required quorum, the meeting could not vote.”. They are not two equally faithful stylistic options under the stated brief. The former introduces a relationship or implication that needs independent justification; the latter retains the intended boundary. Inability to vote is not identical to cancellation of the gathering. Test the exact meaning of the linking expression, not merely the dictionary meanings of the nouns and verbs. Ask if the target asserts causation, expectation violation, permission, sequence, evidential proof, or exclusivity. If it does, locate the corresponding source commitment.

Where the real document is genuinely ambiguous, there are three defensible editorial routes. Preserve the ambiguity through natural target phrasing, seek a qualified source-side answer, or use an approved translator note that separates explanation from the translated claim. Do not silently invent a causal explanation to make the paragraph feel complete. The specific competing possibility here is: Some organizations permit discussion even when binding decisions are impossible. That possibility may or may not be correct, but its continued logical availability shows why an overconfident connective can be dangerous.

Case 8C. Reader-effect and risk diagnostic

A relation error is not just a linguistic blemish; it changes what a reader is licensed to conclude. In this case: Participants might believe presentations and discussion were invalid or absent. Review the target as a person relying on the document rather than as an editor looking for elegant wording. Which action, inference, expectation or accusation becomes more likely after the bad version? What would the intended audience believe if the first source proposition were true but the inferred causal or concessive link were false?

A practical severity review should record the precise sentence, the changed inference, the intended reader, and the possible consequence. If the sentence affects safety, rights, access, technical diagnosis, consent or financial decisions, an unsupported connective deserves more scrutiny than an ordinary style improvement. This scenario’s genre is meeting record, and the source meaning brief is A rule-based dependency is confirmed; discussion may still occur. That is the benchmark for both the accuracy review and the final plain-language edit.

Case 8D. Evidence and author-query gate

An efficient reviewer does not write ‘unclear’ and stop. Ask a question that can be answered with evidence: Did the record state that discussion continued without a vote? Include the source passage and the two candidate target readings. An author can then confirm the relation, provide missing context or explicitly leave the ambiguity open. The record should distinguish an author-confirmed meaning from a translator’s plausible interpretation and from a machine-generated suggestion.

After the answer, update the relation map, not only the visible sentence. If similar segments occur elsewhere, check them for consistency without mechanically propagating the same connective. Context may legitimately make the same pair of vocabulary items express a different relationship in a later document. For this exercise, the brief already supplies the acceptable decision: A rule-based dependency is confirmed; discussion may still occur. The next step is to verify that the chosen target preserves it and creates no new attribution.

Case 8E. Adversarial paraphrase and back-reading

Take the faithful candidate — “Without the required quorum, the meeting could not vote.” — and paraphrase it into a simple reader belief. Now take the unsafe candidate — “The meeting was cancelled and no business took place.” — and do the same. Write down the extra proposition, presupposition or causal claim that appears in the second version. Inability to vote is not identical to cancellation of the gathering. This comparison is stronger than asking whether both sentences sound natural. A fluent sentence can be misleading, and a back-translation can repeat the same mistake if the reviewer already accepts its assumptions.

For an adversarial check, imagine a reader who accepts the two reported events but rejects the suggested link. Would that reader still be entitled to believe the source? Consider the possibility that some organizations permit discussion even when binding decisions are impossible. If yes, the stronger reading needs explicit textual or authorial support. The test does not require mathematical proof. It disciplines translation by making the asserted relationship visible and by preventing an intuition from hiding inside a connective.

Case 8F. Transfer drill and acceptance criterion

Transfer the same reasoning to another language, medium or genre without carrying over English word order. Keep the propositional facts and the stated relation fixed, but choose a natural construction suitable for the target audience. The final acceptable solution does not have to use the English connector in “Without the required quorum, the meeting could not vote.” literally. It does have to preserve a rule-based dependency is confirmed; discussion may still occur. It must also avoid the new belief introduced by “The meeting was cancelled and no business took place.”.

The review is passed when a target reader can identify the two facts, draw only the source-authorized inference, and answer the practical question raised by this meeting record text without adopting an unsupported explanation. The specific failure warning is: Participants might believe presentations and discussion were invalid or absent. Keep the approval record with this sentence’s context, the reviewer’s decision and any author response. If a future revision changes the source relation, rerun this check rather than treating the existing translation as automatically safe.

Case 09: Urgent safety inspection — Work order

Constructed source meaning: A crack was reported. An inspection was requested.

Intended meaning supplied by the exercise brief: The source supports an administrative response, not a confirmed structural danger. Candidate that fails the brief: “The structure was unsafe, so inspectors closed it.” Defensible rendering under the brief: “After a crack was reported, an inspection was requested.”

Case 9A. Proposition isolation: what is actually stated

In this work order example, begin with the two source propositions without merging them into a single story: A crack was reported. An inspection was requested. The first analytical task is to preserve each proposition and its evidential status. The passage is not a licence to invent a hidden third event, an extra agent or an unreported decision. Reported signs justify checking but do not prove a safety classification. A translator who starts by choosing a smooth connective may overlook that the connective itself carries a claim. Start by marking A and B, their subjects and their temporal or logical scope; then describe the relationship separately in an interpretation note. The relevant relation in the approved training brief is: The source supports an administrative response, not a confirmed structural danger.

A useful minimality test is to ask whether each standalone sentence remains true if the proposed relation is false. The crack report may eventually be judged superficial. That alternative demonstrates why the words on the page and the link between them require separate auditing. Only the brief for this constructed case resolves the intended link. In an actual translation, identify what language, adjacent document context or author confirmation performs the equivalent role. Do not claim certainty merely because an explanation seems intuitive.

Case 9B. Competing reading and explicitness threshold

Compare the unsafe target, “The structure was unsafe, so inspectors closed it.”, with “After a crack was reported, an inspection was requested.”. They are not two equally faithful stylistic options under the stated brief. The former introduces a relationship or implication that needs independent justification; the latter retains the intended boundary. Reported signs justify checking but do not prove a safety classification. Test the exact meaning of the linking expression, not merely the dictionary meanings of the nouns and verbs. Ask if the target asserts causation, expectation violation, permission, sequence, evidential proof, or exclusivity. If it does, locate the corresponding source commitment.

Where the real document is genuinely ambiguous, there are three defensible editorial routes. Preserve the ambiguity through natural target phrasing, seek a qualified source-side answer, or use an approved translator note that separates explanation from the translated claim. Do not silently invent a causal explanation to make the paragraph feel complete. The specific competing possibility here is: The crack report may eventually be judged superficial. That possibility may or may not be correct, but its continued logical availability shows why an overconfident connective can be dangerous.

Case 9C. Reader-effect and risk diagnostic

A relation error is not just a linguistic blemish; it changes what a reader is licensed to conclude. In this case: Readers may infer a closure or confirmed hazard that did not occur. Review the target as a person relying on the document rather than as an editor looking for elegant wording. Which action, inference, expectation or accusation becomes more likely after the bad version? What would the intended audience believe if the first source proposition were true but the inferred causal or concessive link were false?

A practical severity review should record the precise sentence, the changed inference, the intended reader, and the possible consequence. If the sentence affects safety, rights, access, technical diagnosis, consent or financial decisions, an unsupported connective deserves more scrutiny than an ordinary style improvement. This scenario’s genre is work order, and the source meaning brief is The source supports an administrative response, not a confirmed structural danger. That is the benchmark for both the accuracy review and the final plain-language edit.

Case 9D. Evidence and author-query gate

An efficient reviewer does not write ‘unclear’ and stop. Ask a question that can be answered with evidence: Was a closure ordered, or was only an inspection requested? Include the source passage and the two candidate target readings. An author can then confirm the relation, provide missing context or explicitly leave the ambiguity open. The record should distinguish an author-confirmed meaning from a translator’s plausible interpretation and from a machine-generated suggestion.

After the answer, update the relation map, not only the visible sentence. If similar segments occur elsewhere, check them for consistency without mechanically propagating the same connective. Context may legitimately make the same pair of vocabulary items express a different relationship in a later document. For this exercise, the brief already supplies the acceptable decision: The source supports an administrative response, not a confirmed structural danger. The next step is to verify that the chosen target preserves it and creates no new attribution.

Case 9E. Adversarial paraphrase and back-reading

Take the faithful candidate — “After a crack was reported, an inspection was requested.” — and paraphrase it into a simple reader belief. Now take the unsafe candidate — “The structure was unsafe, so inspectors closed it.” — and do the same. Write down the extra proposition, presupposition or causal claim that appears in the second version. Reported signs justify checking but do not prove a safety classification. This comparison is stronger than asking whether both sentences sound natural. A fluent sentence can be misleading, and a back-translation can repeat the same mistake if the reviewer already accepts its assumptions.

For an adversarial check, imagine a reader who accepts the two reported events but rejects the suggested link. Would that reader still be entitled to believe the source? Consider the possibility that the crack report may eventually be judged superficial. If yes, the stronger reading needs explicit textual or authorial support. The test does not require mathematical proof. It disciplines translation by making the asserted relationship visible and by preventing an intuition from hiding inside a connective.

Case 9F. Transfer drill and acceptance criterion

Transfer the same reasoning to another language, medium or genre without carrying over English word order. Keep the propositional facts and the stated relation fixed, but choose a natural construction suitable for the target audience. The final acceptable solution does not have to use the English connector in “After a crack was reported, an inspection was requested.” literally. It does have to preserve the source supports an administrative response, not a confirmed structural danger. It must also avoid the new belief introduced by “The structure was unsafe, so inspectors closed it.”.

The review is passed when a target reader can identify the two facts, draw only the source-authorized inference, and answer the practical question raised by this work order text without adopting an unsupported explanation. The specific failure warning is: Readers may infer a closure or confirmed hazard that did not occur. Keep the approval record with this sentence’s context, the reviewer’s decision and any author response. If a future revision changes the source relation, rerun this check rather than treating the existing translation as automatically safe.

Case 10: Witness with limited sight — Interview transcript

Constructed source meaning: The witness heard footsteps. The witness did not see a face.

Intended meaning supplied by the exercise brief: The relationship is a limitation on identification, not evidence of who moved. Candidate that fails the brief: “The witness identified the person by their footsteps.” Defensible rendering under the brief: “The witness heard footsteps but could not identify a face visually.”

Case 10A. Proposition isolation: what is actually stated

In this interview transcript example, begin with the two source propositions without merging them into a single story: The witness heard footsteps. The witness did not see a face. The first analytical task is to preserve each proposition and its evidential status. The passage is not a licence to invent a hidden third event, an extra agent or an unreported decision. Hearing a sound is not the same as recognizing a person. A translator who starts by choosing a smooth connective may overlook that the connective itself carries a claim. Start by marking A and B, their subjects and their temporal or logical scope; then describe the relationship separately in an interpretation note. The relevant relation in the approved training brief is: The relationship is a limitation on identification, not evidence of who moved.

A useful minimality test is to ask whether each standalone sentence remains true if the proposed relation is false. Someone other than the accused could have made the noise. That alternative demonstrates why the words on the page and the link between them require separate auditing. Only the brief for this constructed case resolves the intended link. In an actual translation, identify what language, adjacent document context or author confirmation performs the equivalent role. Do not claim certainty merely because an explanation seems intuitive.

Case 10B. Competing reading and explicitness threshold

Compare the unsafe target, “The witness identified the person by their footsteps.”, with “The witness heard footsteps but could not identify a face visually.”. They are not two equally faithful stylistic options under the stated brief. The former introduces a relationship or implication that needs independent justification; the latter retains the intended boundary. Hearing a sound is not the same as recognizing a person. Test the exact meaning of the linking expression, not merely the dictionary meanings of the nouns and verbs. Ask if the target asserts causation, expectation violation, permission, sequence, evidential proof, or exclusivity. If it does, locate the corresponding source commitment.

Where the real document is genuinely ambiguous, there are three defensible editorial routes. Preserve the ambiguity through natural target phrasing, seek a qualified source-side answer, or use an approved translator note that separates explanation from the translated claim. Do not silently invent a causal explanation to make the paragraph feel complete. The specific competing possibility here is: Someone other than the accused could have made the noise. That possibility may or may not be correct, but its continued logical availability shows why an overconfident connective can be dangerous.

Case 10C. Reader-effect and risk diagnostic

A relation error is not just a linguistic blemish; it changes what a reader is licensed to conclude. In this case: A legal reader could assign certainty to an unverified identification. Review the target as a person relying on the document rather than as an editor looking for elegant wording. Which action, inference, expectation or accusation becomes more likely after the bad version? What would the intended audience believe if the first source proposition were true but the inferred causal or concessive link were false?

A practical severity review should record the precise sentence, the changed inference, the intended reader, and the possible consequence. If the sentence affects safety, rights, access, technical diagnosis, consent or financial decisions, an unsupported connective deserves more scrutiny than an ordinary style improvement. This scenario’s genre is interview transcript, and the source meaning brief is The relationship is a limitation on identification, not evidence of who moved. That is the benchmark for both the accuracy review and the final plain-language edit.

Case 10D. Evidence and author-query gate

An efficient reviewer does not write ‘unclear’ and stop. Ask a question that can be answered with evidence: Did the witness claim to recognize a gait or only hear footsteps? Include the source passage and the two candidate target readings. An author can then confirm the relation, provide missing context or explicitly leave the ambiguity open. The record should distinguish an author-confirmed meaning from a translator’s plausible interpretation and from a machine-generated suggestion.

After the answer, update the relation map, not only the visible sentence. If similar segments occur elsewhere, check them for consistency without mechanically propagating the same connective. Context may legitimately make the same pair of vocabulary items express a different relationship in a later document. For this exercise, the brief already supplies the acceptable decision: The relationship is a limitation on identification, not evidence of who moved. The next step is to verify that the chosen target preserves it and creates no new attribution.

Case 10E. Adversarial paraphrase and back-reading

Take the faithful candidate — “The witness heard footsteps but could not identify a face visually.” — and paraphrase it into a simple reader belief. Now take the unsafe candidate — “The witness identified the person by their footsteps.” — and do the same. Write down the extra proposition, presupposition or causal claim that appears in the second version. Hearing a sound is not the same as recognizing a person. This comparison is stronger than asking whether both sentences sound natural. A fluent sentence can be misleading, and a back-translation can repeat the same mistake if the reviewer already accepts its assumptions.

For an adversarial check, imagine a reader who accepts the two reported events but rejects the suggested link. Would that reader still be entitled to believe the source? Consider the possibility that someone other than the accused could have made the noise. If yes, the stronger reading needs explicit textual or authorial support. The test does not require mathematical proof. It disciplines translation by making the asserted relationship visible and by preventing an intuition from hiding inside a connective.

Case 10F. Transfer drill and acceptance criterion

Transfer the same reasoning to another language, medium or genre without carrying over English word order. Keep the propositional facts and the stated relation fixed, but choose a natural construction suitable for the target audience. The final acceptable solution does not have to use the English connector in “The witness heard footsteps but could not identify a face visually.” literally. It does have to preserve the relationship is a limitation on identification, not evidence of who moved. It must also avoid the new belief introduced by “The witness identified the person by their footsteps.”.

The review is passed when a target reader can identify the two facts, draw only the source-authorized inference, and answer the practical question raised by this interview transcript text without adopting an unsupported explanation. The specific failure warning is: A legal reader could assign certainty to an unverified identification. Keep the approval record with this sentence’s context, the reviewer’s decision and any author response. If a future revision changes the source relation, rerun this check rather than treating the existing translation as automatically safe.

Case 11: Notification settings — Software help

Constructed source meaning: Notifications were muted. The vibration alert continued.

Intended meaning supplied by the exercise brief: Two notification channels have different control settings; a contrast is intended. Candidate that fails the brief: “Vibration stopped as a result of muting notifications.” Defensible rendering under the brief: “Notifications were muted, but vibration alerts continued.”

Case 11A. Proposition isolation: what is actually stated

In this software help example, begin with the two source propositions without merging them into a single story: Notifications were muted. The vibration alert continued. The first analytical task is to preserve each proposition and its evidential status. The passage is not a licence to invent a hidden third event, an extra agent or an unreported decision. The target must not collapse sound notifications and haptic alerts into one feature. A translator who starts by choosing a smooth connective may overlook that the connective itself carries a claim. Start by marking A and B, their subjects and their temporal or logical scope; then describe the relationship separately in an interpretation note. The relevant relation in the approved training brief is: Two notification channels have different control settings; a contrast is intended.

A useful minimality test is to ask whether each standalone sentence remains true if the proposed relation is false. A later software version could change the interface. That alternative demonstrates why the words on the page and the link between them require separate auditing. Only the brief for this constructed case resolves the intended link. In an actual translation, identify what language, adjacent document context or author confirmation performs the equivalent role. Do not claim certainty merely because an explanation seems intuitive.

Case 11B. Competing reading and explicitness threshold

Compare the unsafe target, “Vibration stopped as a result of muting notifications.”, with “Notifications were muted, but vibration alerts continued.”. They are not two equally faithful stylistic options under the stated brief. The former introduces a relationship or implication that needs independent justification; the latter retains the intended boundary. The target must not collapse sound notifications and haptic alerts into one feature. Test the exact meaning of the linking expression, not merely the dictionary meanings of the nouns and verbs. Ask if the target asserts causation, expectation violation, permission, sequence, evidential proof, or exclusivity. If it does, locate the corresponding source commitment.

Where the real document is genuinely ambiguous, there are three defensible editorial routes. Preserve the ambiguity through natural target phrasing, seek a qualified source-side answer, or use an approved translator note that separates explanation from the translated claim. Do not silently invent a causal explanation to make the paragraph feel complete. The specific competing possibility here is: A later software version could change the interface. That possibility may or may not be correct, but its continued logical availability shows why an overconfident connective can be dangerous.

Case 11C. Reader-effect and risk diagnostic

A relation error is not just a linguistic blemish; it changes what a reader is licensed to conclude. In this case: Users may assume they will not be disturbed. Review the target as a person relying on the document rather than as an editor looking for elegant wording. Which action, inference, expectation or accusation becomes more likely after the bad version? What would the intended audience believe if the first source proposition were true but the inferred causal or concessive link were false?

A practical severity review should record the precise sentence, the changed inference, the intended reader, and the possible consequence. If the sentence affects safety, rights, access, technical diagnosis, consent or financial decisions, an unsupported connective deserves more scrutiny than an ordinary style improvement. This scenario’s genre is software help, and the source meaning brief is Two notification channels have different control settings; a contrast is intended. That is the benchmark for both the accuracy review and the final plain-language edit.

Case 11D. Evidence and author-query gate

An efficient reviewer does not write ‘unclear’ and stop. Ask a question that can be answered with evidence: Are visual badges, vibration and sounds controlled by separate toggles? Include the source passage and the two candidate target readings. An author can then confirm the relation, provide missing context or explicitly leave the ambiguity open. The record should distinguish an author-confirmed meaning from a translator’s plausible interpretation and from a machine-generated suggestion.

After the answer, update the relation map, not only the visible sentence. If similar segments occur elsewhere, check them for consistency without mechanically propagating the same connective. Context may legitimately make the same pair of vocabulary items express a different relationship in a later document. For this exercise, the brief already supplies the acceptable decision: Two notification channels have different control settings; a contrast is intended. The next step is to verify that the chosen target preserves it and creates no new attribution.

Case 11E. Adversarial paraphrase and back-reading

Take the faithful candidate — “Notifications were muted, but vibration alerts continued.” — and paraphrase it into a simple reader belief. Now take the unsafe candidate — “Vibration stopped as a result of muting notifications.” — and do the same. Write down the extra proposition, presupposition or causal claim that appears in the second version. The target must not collapse sound notifications and haptic alerts into one feature. This comparison is stronger than asking whether both sentences sound natural. A fluent sentence can be misleading, and a back-translation can repeat the same mistake if the reviewer already accepts its assumptions.

For an adversarial check, imagine a reader who accepts the two reported events but rejects the suggested link. Would that reader still be entitled to believe the source? Consider the possibility that a later software version could change the interface. If yes, the stronger reading needs explicit textual or authorial support. The test does not require mathematical proof. It disciplines translation by making the asserted relationship visible and by preventing an intuition from hiding inside a connective.

Case 11F. Transfer drill and acceptance criterion

Transfer the same reasoning to another language, medium or genre without carrying over English word order. Keep the propositional facts and the stated relation fixed, but choose a natural construction suitable for the target audience. The final acceptable solution does not have to use the English connector in “Notifications were muted, but vibration alerts continued.” literally. It does have to preserve two notification channels have different control settings; a contrast is intended. It must also avoid the new belief introduced by “Vibration stopped as a result of muting notifications.”.

The review is passed when a target reader can identify the two facts, draw only the source-authorized inference, and answer the practical question raised by this software help text without adopting an unsupported explanation. The specific failure warning is: Users may assume they will not be disturbed. Keep the approval record with this sentence’s context, the reviewer’s decision and any author response. If a future revision changes the source relation, rerun this check rather than treating the existing translation as automatically safe.

Case 12: Signed document and scan — Records note

Constructed source meaning: The form was signed. A scan was uploaded.

Intended meaning supplied by the exercise brief: A sequence of document handling steps is described without asserting legal effect. Candidate that fails the brief: “Uploading the scan legally approved the application.” Defensible rendering under the brief: “The form was signed, and a scan was uploaded afterward.”

Case 12A. Proposition isolation: what is actually stated

In this records note example, begin with the two source propositions without merging them into a single story: The form was signed. A scan was uploaded. The first analytical task is to preserve each proposition and its evidential status. The passage is not a licence to invent a hidden third event, an extra agent or an unreported decision. The wrong translation invents official acceptance or legal validation. A translator who starts by choosing a smooth connective may overlook that the connective itself carries a claim. Start by marking A and B, their subjects and their temporal or logical scope; then describe the relationship separately in an interpretation note. The relevant relation in the approved training brief is: A sequence of document handling steps is described without asserting legal effect.

A useful minimality test is to ask whether each standalone sentence remains true if the proposed relation is false. A submitted scan may later be rejected for legibility. That alternative demonstrates why the words on the page and the link between them require separate auditing. Only the brief for this constructed case resolves the intended link. In an actual translation, identify what language, adjacent document context or author confirmation performs the equivalent role. Do not claim certainty merely because an explanation seems intuitive.

Case 12B. Competing reading and explicitness threshold

Compare the unsafe target, “Uploading the scan legally approved the application.”, with “The form was signed, and a scan was uploaded afterward.”. They are not two equally faithful stylistic options under the stated brief. The former introduces a relationship or implication that needs independent justification; the latter retains the intended boundary. The wrong translation invents official acceptance or legal validation. Test the exact meaning of the linking expression, not merely the dictionary meanings of the nouns and verbs. Ask if the target asserts causation, expectation violation, permission, sequence, evidential proof, or exclusivity. If it does, locate the corresponding source commitment.

Where the real document is genuinely ambiguous, there are three defensible editorial routes. Preserve the ambiguity through natural target phrasing, seek a qualified source-side answer, or use an approved translator note that separates explanation from the translated claim. Do not silently invent a causal explanation to make the paragraph feel complete. The specific competing possibility here is: A submitted scan may later be rejected for legibility. That possibility may or may not be correct, but its continued logical availability shows why an overconfident connective can be dangerous.

Case 12C. Reader-effect and risk diagnostic

A relation error is not just a linguistic blemish; it changes what a reader is licensed to conclude. In this case: Applicants may stop required follow-up steps. Review the target as a person relying on the document rather than as an editor looking for elegant wording. Which action, inference, expectation or accusation becomes more likely after the bad version? What would the intended audience believe if the first source proposition were true but the inferred causal or concessive link were false?

A practical severity review should record the precise sentence, the changed inference, the intended reader, and the possible consequence. If the sentence affects safety, rights, access, technical diagnosis, consent or financial decisions, an unsupported connective deserves more scrutiny than an ordinary style improvement. This scenario’s genre is records note, and the source meaning brief is A sequence of document handling steps is described without asserting legal effect. That is the benchmark for both the accuracy review and the final plain-language edit.

Case 12D. Evidence and author-query gate

An efficient reviewer does not write ‘unclear’ and stop. Ask a question that can be answered with evidence: Was formal acceptance recorded anywhere or only a document upload? Include the source passage and the two candidate target readings. An author can then confirm the relation, provide missing context or explicitly leave the ambiguity open. The record should distinguish an author-confirmed meaning from a translator’s plausible interpretation and from a machine-generated suggestion.

After the answer, update the relation map, not only the visible sentence. If similar segments occur elsewhere, check them for consistency without mechanically propagating the same connective. Context may legitimately make the same pair of vocabulary items express a different relationship in a later document. For this exercise, the brief already supplies the acceptable decision: A sequence of document handling steps is described without asserting legal effect. The next step is to verify that the chosen target preserves it and creates no new attribution.

Case 12E. Adversarial paraphrase and back-reading

Take the faithful candidate — “The form was signed, and a scan was uploaded afterward.” — and paraphrase it into a simple reader belief. Now take the unsafe candidate — “Uploading the scan legally approved the application.” — and do the same. Write down the extra proposition, presupposition or causal claim that appears in the second version. The wrong translation invents official acceptance or legal validation. This comparison is stronger than asking whether both sentences sound natural. A fluent sentence can be misleading, and a back-translation can repeat the same mistake if the reviewer already accepts its assumptions.

For an adversarial check, imagine a reader who accepts the two reported events but rejects the suggested link. Would that reader still be entitled to believe the source? Consider the possibility that a submitted scan may later be rejected for legibility. If yes, the stronger reading needs explicit textual or authorial support. The test does not require mathematical proof. It disciplines translation by making the asserted relationship visible and by preventing an intuition from hiding inside a connective.

Case 12F. Transfer drill and acceptance criterion

Transfer the same reasoning to another language, medium or genre without carrying over English word order. Keep the propositional facts and the stated relation fixed, but choose a natural construction suitable for the target audience. The final acceptable solution does not have to use the English connector in “The form was signed, and a scan was uploaded afterward.” literally. It does have to preserve a sequence of document handling steps is described without asserting legal effect. It must also avoid the new belief introduced by “Uploading the scan legally approved the application.”.

The review is passed when a target reader can identify the two facts, draw only the source-authorized inference, and answer the practical question raised by this records note text without adopting an unsupported explanation. The specific failure warning is: Applicants may stop required follow-up steps. Keep the approval record with this sentence’s context, the reviewer’s decision and any author response. If a future revision changes the source relation, rerun this check rather than treating the existing translation as automatically safe.

Case 13: Museum label and attribution — Exhibition label

Constructed source meaning: A painting was displayed beside the sketch. The sketch may have preceded it.

Intended meaning supplied by the exercise brief: Physical juxtaposition is certain; chronology is hedged. Candidate that fails the brief: “The sketch definitively caused the final painting.” Defensible rendering under the brief: “The painting hangs beside a sketch that may have been made earlier.”

Case 13A. Proposition isolation: what is actually stated

In this exhibition label example, begin with the two source propositions without merging them into a single story: A painting was displayed beside the sketch. The sketch may have preceded it. The first analytical task is to preserve each proposition and its evidential status. The passage is not a licence to invent a hidden third event, an extra agent or an unreported decision. The proposed chronology is not established provenance or influence. A translator who starts by choosing a smooth connective may overlook that the connective itself carries a claim. Start by marking A and B, their subjects and their temporal or logical scope; then describe the relationship separately in an interpretation note. The relevant relation in the approved training brief is: Physical juxtaposition is certain; chronology is hedged.

A useful minimality test is to ask whether each standalone sentence remains true if the proposed relation is false. The sketch could represent later study of the painting. That alternative demonstrates why the words on the page and the link between them require separate auditing. Only the brief for this constructed case resolves the intended link. In an actual translation, identify what language, adjacent document context or author confirmation performs the equivalent role. Do not claim certainty merely because an explanation seems intuitive.

Case 13B. Competing reading and explicitness threshold

Compare the unsafe target, “The sketch definitively caused the final painting.”, with “The painting hangs beside a sketch that may have been made earlier.”. They are not two equally faithful stylistic options under the stated brief. The former introduces a relationship or implication that needs independent justification; the latter retains the intended boundary. The proposed chronology is not established provenance or influence. Test the exact meaning of the linking expression, not merely the dictionary meanings of the nouns and verbs. Ask if the target asserts causation, expectation violation, permission, sequence, evidential proof, or exclusivity. If it does, locate the corresponding source commitment.

Where the real document is genuinely ambiguous, there are three defensible editorial routes. Preserve the ambiguity through natural target phrasing, seek a qualified source-side answer, or use an approved translator note that separates explanation from the translated claim. Do not silently invent a causal explanation to make the paragraph feel complete. The specific competing possibility here is: The sketch could represent later study of the painting. That possibility may or may not be correct, but its continued logical availability shows why an overconfident connective can be dangerous.

Case 13C. Reader-effect and risk diagnostic

A relation error is not just a linguistic blemish; it changes what a reader is licensed to conclude. In this case: Visitors may mistake an interpretive possibility for a curatorial finding. Review the target as a person relying on the document rather than as an editor looking for elegant wording. Which action, inference, expectation or accusation becomes more likely after the bad version? What would the intended audience believe if the first source proposition were true but the inferred causal or concessive link were false?

A practical severity review should record the precise sentence, the changed inference, the intended reader, and the possible consequence. If the sentence affects safety, rights, access, technical diagnosis, consent or financial decisions, an unsupported connective deserves more scrutiny than an ordinary style improvement. This scenario’s genre is exhibition label, and the source meaning brief is Physical juxtaposition is certain; chronology is hedged. That is the benchmark for both the accuracy review and the final plain-language edit.

Case 13D. Evidence and author-query gate

An efficient reviewer does not write ‘unclear’ and stop. Ask a question that can be answered with evidence: Is there dated documentary evidence for the sketch’s order? Include the source passage and the two candidate target readings. An author can then confirm the relation, provide missing context or explicitly leave the ambiguity open. The record should distinguish an author-confirmed meaning from a translator’s plausible interpretation and from a machine-generated suggestion.

After the answer, update the relation map, not only the visible sentence. If similar segments occur elsewhere, check them for consistency without mechanically propagating the same connective. Context may legitimately make the same pair of vocabulary items express a different relationship in a later document. For this exercise, the brief already supplies the acceptable decision: Physical juxtaposition is certain; chronology is hedged. The next step is to verify that the chosen target preserves it and creates no new attribution.

Case 13E. Adversarial paraphrase and back-reading

Take the faithful candidate — “The painting hangs beside a sketch that may have been made earlier.” — and paraphrase it into a simple reader belief. Now take the unsafe candidate — “The sketch definitively caused the final painting.” — and do the same. Write down the extra proposition, presupposition or causal claim that appears in the second version. The proposed chronology is not established provenance or influence. This comparison is stronger than asking whether both sentences sound natural. A fluent sentence can be misleading, and a back-translation can repeat the same mistake if the reviewer already accepts its assumptions.

For an adversarial check, imagine a reader who accepts the two reported events but rejects the suggested link. Would that reader still be entitled to believe the source? Consider the possibility that the sketch could represent later study of the painting. If yes, the stronger reading needs explicit textual or authorial support. The test does not require mathematical proof. It disciplines translation by making the asserted relationship visible and by preventing an intuition from hiding inside a connective.

Case 13F. Transfer drill and acceptance criterion

Transfer the same reasoning to another language, medium or genre without carrying over English word order. Keep the propositional facts and the stated relation fixed, but choose a natural construction suitable for the target audience. The final acceptable solution does not have to use the English connector in “The painting hangs beside a sketch that may have been made earlier.” literally. It does have to preserve physical juxtaposition is certain; chronology is hedged. It must also avoid the new belief introduced by “The sketch definitively caused the final painting.”.

The review is passed when a target reader can identify the two facts, draw only the source-authorized inference, and answer the practical question raised by this exhibition label text without adopting an unsupported explanation. The specific failure warning is: Visitors may mistake an interpretive possibility for a curatorial finding. Keep the approval record with this sentence’s context, the reviewer’s decision and any author response. If a future revision changes the source relation, rerun this check rather than treating the existing translation as automatically safe.

Case 14: Public transport detour — Travel notice

Constructed source meaning: The main crossing was unavailable. Buses used an alternative route.

Intended meaning supplied by the exercise brief: The official brief attributes the detour to the crossing closure. Candidate that fails the brief: “The crossing was closed after buses changed routes.” Defensible rendering under the brief: “Because the main crossing was unavailable, buses used an alternative route.”

Case 14A. Proposition isolation: what is actually stated

In this travel notice example, begin with the two source propositions without merging them into a single story: The main crossing was unavailable. Buses used an alternative route. The first analytical task is to preserve each proposition and its evidential status. The passage is not a licence to invent a hidden third event, an extra agent or an unreported decision. The faulty ordering reverses the causal dependency and may obscure the route restriction. A translator who starts by choosing a smooth connective may overlook that the connective itself carries a claim. Start by marking A and B, their subjects and their temporal or logical scope; then describe the relationship separately in an interpretation note. The relevant relation in the approved training brief is: The official brief attributes the detour to the crossing closure.

A useful minimality test is to ask whether each standalone sentence remains true if the proposed relation is false. Another obstruction could also have affected buses. That alternative demonstrates why the words on the page and the link between them require separate auditing. Only the brief for this constructed case resolves the intended link. In an actual translation, identify what language, adjacent document context or author confirmation performs the equivalent role. Do not claim certainty merely because an explanation seems intuitive.

Case 14B. Competing reading and explicitness threshold

Compare the unsafe target, “The crossing was closed after buses changed routes.”, with “Because the main crossing was unavailable, buses used an alternative route.”. They are not two equally faithful stylistic options under the stated brief. The former introduces a relationship or implication that needs independent justification; the latter retains the intended boundary. The faulty ordering reverses the causal dependency and may obscure the route restriction. Test the exact meaning of the linking expression, not merely the dictionary meanings of the nouns and verbs. Ask if the target asserts causation, expectation violation, permission, sequence, evidential proof, or exclusivity. If it does, locate the corresponding source commitment.

Where the real document is genuinely ambiguous, there are three defensible editorial routes. Preserve the ambiguity through natural target phrasing, seek a qualified source-side answer, or use an approved translator note that separates explanation from the translated claim. Do not silently invent a causal explanation to make the paragraph feel complete. The specific competing possibility here is: Another obstruction could also have affected buses. That possibility may or may not be correct, but its continued logical availability shows why an overconfident connective can be dangerous.

Case 14C. Reader-effect and risk diagnostic

A relation error is not just a linguistic blemish; it changes what a reader is licensed to conclude. In this case: Travelers could infer the route diversion caused the closure. Review the target as a person relying on the document rather than as an editor looking for elegant wording. Which action, inference, expectation or accusation becomes more likely after the bad version? What would the intended audience believe if the first source proposition were true but the inferred causal or concessive link were false?

A practical severity review should record the precise sentence, the changed inference, the intended reader, and the possible consequence. If the sentence affects safety, rights, access, technical diagnosis, consent or financial decisions, an unsupported connective deserves more scrutiny than an ordinary style improvement. This scenario’s genre is travel notice, and the source meaning brief is The official brief attributes the detour to the crossing closure. That is the benchmark for both the accuracy review and the final plain-language edit.

Case 14D. Evidence and author-query gate

An efficient reviewer does not write ‘unclear’ and stop. Ask a question that can be answered with evidence: Was the crossing closure the sole cause of the route change? Include the source passage and the two candidate target readings. An author can then confirm the relation, provide missing context or explicitly leave the ambiguity open. The record should distinguish an author-confirmed meaning from a translator’s plausible interpretation and from a machine-generated suggestion.

After the answer, update the relation map, not only the visible sentence. If similar segments occur elsewhere, check them for consistency without mechanically propagating the same connective. Context may legitimately make the same pair of vocabulary items express a different relationship in a later document. For this exercise, the brief already supplies the acceptable decision: The official brief attributes the detour to the crossing closure. The next step is to verify that the chosen target preserves it and creates no new attribution.

Case 14E. Adversarial paraphrase and back-reading

Take the faithful candidate — “Because the main crossing was unavailable, buses used an alternative route.” — and paraphrase it into a simple reader belief. Now take the unsafe candidate — “The crossing was closed after buses changed routes.” — and do the same. Write down the extra proposition, presupposition or causal claim that appears in the second version. The faulty ordering reverses the causal dependency and may obscure the route restriction. This comparison is stronger than asking whether both sentences sound natural. A fluent sentence can be misleading, and a back-translation can repeat the same mistake if the reviewer already accepts its assumptions.

For an adversarial check, imagine a reader who accepts the two reported events but rejects the suggested link. Would that reader still be entitled to believe the source? Consider the possibility that another obstruction could also have affected buses. If yes, the stronger reading needs explicit textual or authorial support. The test does not require mathematical proof. It disciplines translation by making the asserted relationship visible and by preventing an intuition from hiding inside a connective.

Case 14F. Transfer drill and acceptance criterion

Transfer the same reasoning to another language, medium or genre without carrying over English word order. Keep the propositional facts and the stated relation fixed, but choose a natural construction suitable for the target audience. The final acceptable solution does not have to use the English connector in “Because the main crossing was unavailable, buses used an alternative route.” literally. It does have to preserve the official brief attributes the detour to the crossing closure. It must also avoid the new belief introduced by “The crossing was closed after buses changed routes.”.

The review is passed when a target reader can identify the two facts, draw only the source-authorized inference, and answer the practical question raised by this travel notice text without adopting an unsupported explanation. The specific failure warning is: Travelers could infer the route diversion caused the closure. Keep the approval record with this sentence’s context, the reviewer’s decision and any author response. If a future revision changes the source relation, rerun this check rather than treating the existing translation as automatically safe.

Case 15: Processing trigger — Application policy

Constructed source meaning: An application is accepted. Processing begins.

Intended meaning supplied by the exercise brief: Acceptance is a precondition in a general rule, not a promise of acceptance. Candidate that fails the brief: “All applications will be accepted and processed immediately.” Defensible rendering under the brief: “If an application is accepted, processing begins.”

Case 15A. Proposition isolation: what is actually stated

In this application policy example, begin with the two source propositions without merging them into a single story: An application is accepted. Processing begins. The first analytical task is to preserve each proposition and its evidential status. The passage is not a licence to invent a hidden third event, an extra agent or an unreported decision. An if-rule must not be turned into a guaranteed outcome. A translator who starts by choosing a smooth connective may overlook that the connective itself carries a claim. Start by marking A and B, their subjects and their temporal or logical scope; then describe the relationship separately in an interpretation note. The relevant relation in the approved training brief is: Acceptance is a precondition in a general rule, not a promise of acceptance.

A useful minimality test is to ask whether each standalone sentence remains true if the proposed relation is false. Rejection remains a possible outcome. That alternative demonstrates why the words on the page and the link between them require separate auditing. Only the brief for this constructed case resolves the intended link. In an actual translation, identify what language, adjacent document context or author confirmation performs the equivalent role. Do not claim certainty merely because an explanation seems intuitive.

Case 15B. Competing reading and explicitness threshold

Compare the unsafe target, “All applications will be accepted and processed immediately.”, with “If an application is accepted, processing begins.”. They are not two equally faithful stylistic options under the stated brief. The former introduces a relationship or implication that needs independent justification; the latter retains the intended boundary. An if-rule must not be turned into a guaranteed outcome. Test the exact meaning of the linking expression, not merely the dictionary meanings of the nouns and verbs. Ask if the target asserts causation, expectation violation, permission, sequence, evidential proof, or exclusivity. If it does, locate the corresponding source commitment.

Where the real document is genuinely ambiguous, there are three defensible editorial routes. Preserve the ambiguity through natural target phrasing, seek a qualified source-side answer, or use an approved translator note that separates explanation from the translated claim. Do not silently invent a causal explanation to make the paragraph feel complete. The specific competing possibility here is: Rejection remains a possible outcome. That possibility may or may not be correct, but its continued logical availability shows why an overconfident connective can be dangerous.

Case 15C. Reader-effect and risk diagnostic

A relation error is not just a linguistic blemish; it changes what a reader is licensed to conclude. In this case: Applicants could rely on processing that has not been authorized. Review the target as a person relying on the document rather than as an editor looking for elegant wording. Which action, inference, expectation or accusation becomes more likely after the bad version? What would the intended audience believe if the first source proposition were true but the inferred causal or concessive link were false?

A practical severity review should record the precise sentence, the changed inference, the intended reader, and the possible consequence. If the sentence affects safety, rights, access, technical diagnosis, consent or financial decisions, an unsupported connective deserves more scrutiny than an ordinary style improvement. This scenario’s genre is application policy, and the source meaning brief is Acceptance is a precondition in a general rule, not a promise of acceptance. That is the benchmark for both the accuracy review and the final plain-language edit.

Case 15D. Evidence and author-query gate

An efficient reviewer does not write ‘unclear’ and stop. Ask a question that can be answered with evidence: Does ‘processing begins’ specify a timeline or merely a stage transition? Include the source passage and the two candidate target readings. An author can then confirm the relation, provide missing context or explicitly leave the ambiguity open. The record should distinguish an author-confirmed meaning from a translator’s plausible interpretation and from a machine-generated suggestion.

After the answer, update the relation map, not only the visible sentence. If similar segments occur elsewhere, check them for consistency without mechanically propagating the same connective. Context may legitimately make the same pair of vocabulary items express a different relationship in a later document. For this exercise, the brief already supplies the acceptable decision: Acceptance is a precondition in a general rule, not a promise of acceptance. The next step is to verify that the chosen target preserves it and creates no new attribution.

Case 15E. Adversarial paraphrase and back-reading

Take the faithful candidate — “If an application is accepted, processing begins.” — and paraphrase it into a simple reader belief. Now take the unsafe candidate — “All applications will be accepted and processed immediately.” — and do the same. Write down the extra proposition, presupposition or causal claim that appears in the second version. An if-rule must not be turned into a guaranteed outcome. This comparison is stronger than asking whether both sentences sound natural. A fluent sentence can be misleading, and a back-translation can repeat the same mistake if the reviewer already accepts its assumptions.

For an adversarial check, imagine a reader who accepts the two reported events but rejects the suggested link. Would that reader still be entitled to believe the source? Consider the possibility that rejection remains a possible outcome. If yes, the stronger reading needs explicit textual or authorial support. The test does not require mathematical proof. It disciplines translation by making the asserted relationship visible and by preventing an intuition from hiding inside a connective.

Case 15F. Transfer drill and acceptance criterion

Transfer the same reasoning to another language, medium or genre without carrying over English word order. Keep the propositional facts and the stated relation fixed, but choose a natural construction suitable for the target audience. The final acceptable solution does not have to use the English connector in “If an application is accepted, processing begins.” literally. It does have to preserve acceptance is a precondition in a general rule, not a promise of acceptance. It must also avoid the new belief introduced by “All applications will be accepted and processed immediately.”.

The review is passed when a target reader can identify the two facts, draw only the source-authorized inference, and answer the practical question raised by this application policy text without adopting an unsupported explanation. The specific failure warning is: Applicants could rely on processing that has not been authorized. Keep the approval record with this sentence’s context, the reviewer’s decision and any author response. If a future revision changes the source relation, rerun this check rather than treating the existing translation as automatically safe.

Case 16: Limited gallery access — Access advisory

Constructed source meaning: The museum is open. The east gallery is accessible by reservation only.

Intended meaning supplied by the exercise brief: The exception constrains one space within a broader opening statement. Candidate that fails the brief: “Every gallery is open without reservation.” Defensible rendering under the brief: “The museum is open; access to the east gallery requires a reservation.”

Case 16A. Proposition isolation: what is actually stated

In this access advisory example, begin with the two source propositions without merging them into a single story: The museum is open. The east gallery is accessible by reservation only. The first analytical task is to preserve each proposition and its evidential status. The passage is not a licence to invent a hidden third event, an extra agent or an unreported decision. The misleading version removes a consequential exception. A translator who starts by choosing a smooth connective may overlook that the connective itself carries a claim. Start by marking A and B, their subjects and their temporal or logical scope; then describe the relationship separately in an interpretation note. The relevant relation in the approved training brief is: The exception constrains one space within a broader opening statement.

A useful minimality test is to ask whether each standalone sentence remains true if the proposed relation is false. Other areas may be freely accessible. That alternative demonstrates why the words on the page and the link between them require separate auditing. Only the brief for this constructed case resolves the intended link. In an actual translation, identify what language, adjacent document context or author confirmation performs the equivalent role. Do not claim certainty merely because an explanation seems intuitive.

Case 16B. Competing reading and explicitness threshold

Compare the unsafe target, “Every gallery is open without reservation.”, with “The museum is open; access to the east gallery requires a reservation.”. They are not two equally faithful stylistic options under the stated brief. The former introduces a relationship or implication that needs independent justification; the latter retains the intended boundary. The misleading version removes a consequential exception. Test the exact meaning of the linking expression, not merely the dictionary meanings of the nouns and verbs. Ask if the target asserts causation, expectation violation, permission, sequence, evidential proof, or exclusivity. If it does, locate the corresponding source commitment.

Where the real document is genuinely ambiguous, there are three defensible editorial routes. Preserve the ambiguity through natural target phrasing, seek a qualified source-side answer, or use an approved translator note that separates explanation from the translated claim. Do not silently invent a causal explanation to make the paragraph feel complete. The specific competing possibility here is: Other areas may be freely accessible. That possibility may or may not be correct, but its continued logical availability shows why an overconfident connective can be dangerous.

Case 16C. Reader-effect and risk diagnostic

A relation error is not just a linguistic blemish; it changes what a reader is licensed to conclude. In this case: Visitors may reach a restricted area without a booking. Review the target as a person relying on the document rather than as an editor looking for elegant wording. Which action, inference, expectation or accusation becomes more likely after the bad version? What would the intended audience believe if the first source proposition were true but the inferred causal or concessive link were false?

A practical severity review should record the precise sentence, the changed inference, the intended reader, and the possible consequence. If the sentence affects safety, rights, access, technical diagnosis, consent or financial decisions, an unsupported connective deserves more scrutiny than an ordinary style improvement. This scenario’s genre is access advisory, and the source meaning brief is The exception constrains one space within a broader opening statement. That is the benchmark for both the accuracy review and the final plain-language edit.

Case 16D. Evidence and author-query gate

An efficient reviewer does not write ‘unclear’ and stop. Ask a question that can be answered with evidence: Does the reservation rule apply to the east gallery only? Include the source passage and the two candidate target readings. An author can then confirm the relation, provide missing context or explicitly leave the ambiguity open. The record should distinguish an author-confirmed meaning from a translator’s plausible interpretation and from a machine-generated suggestion.

After the answer, update the relation map, not only the visible sentence. If similar segments occur elsewhere, check them for consistency without mechanically propagating the same connective. Context may legitimately make the same pair of vocabulary items express a different relationship in a later document. For this exercise, the brief already supplies the acceptable decision: The exception constrains one space within a broader opening statement. The next step is to verify that the chosen target preserves it and creates no new attribution.

Case 16E. Adversarial paraphrase and back-reading

Take the faithful candidate — “The museum is open; access to the east gallery requires a reservation.” — and paraphrase it into a simple reader belief. Now take the unsafe candidate — “Every gallery is open without reservation.” — and do the same. Write down the extra proposition, presupposition or causal claim that appears in the second version. The misleading version removes a consequential exception. This comparison is stronger than asking whether both sentences sound natural. A fluent sentence can be misleading, and a back-translation can repeat the same mistake if the reviewer already accepts its assumptions.

For an adversarial check, imagine a reader who accepts the two reported events but rejects the suggested link. Would that reader still be entitled to believe the source? Consider the possibility that other areas may be freely accessible. If yes, the stronger reading needs explicit textual or authorial support. The test does not require mathematical proof. It disciplines translation by making the asserted relationship visible and by preventing an intuition from hiding inside a connective.

Case 16F. Transfer drill and acceptance criterion

Transfer the same reasoning to another language, medium or genre without carrying over English word order. Keep the propositional facts and the stated relation fixed, but choose a natural construction suitable for the target audience. The final acceptable solution does not have to use the English connector in “The museum is open; access to the east gallery requires a reservation.” literally. It does have to preserve the exception constrains one space within a broader opening statement. It must also avoid the new belief introduced by “Every gallery is open without reservation.”.

The review is passed when a target reader can identify the two facts, draw only the source-authorized inference, and answer the practical question raised by this access advisory text without adopting an unsupported explanation. The specific failure warning is: Visitors may reach a restricted area without a booking. Keep the approval record with this sentence’s context, the reviewer’s decision and any author response. If a future revision changes the source relation, rerun this check rather than treating the existing translation as automatically safe.

Case 17: Experimental control result — Research results

Constructed source meaning: The intervention group improved. The comparison group also improved.

Intended meaning supplied by the exercise brief: The second observation complicates attribution; no net treatment effect is provided. Candidate that fails the brief: “The treatment caused all the improvement.” Defensible rendering under the brief: “Both groups improved; this comparison alone does not establish the intervention’s effect.”

Case 17A. Proposition isolation: what is actually stated

In this research results example, begin with the two source propositions without merging them into a single story: The intervention group improved. The comparison group also improved. The first analytical task is to preserve each proposition and its evidential status. The passage is not a licence to invent a hidden third event, an extra agent or an unreported decision. A treatment claim requires an appropriate difference and analysis, not just a within-group change. A translator who starts by choosing a smooth connective may overlook that the connective itself carries a claim. Start by marking A and B, their subjects and their temporal or logical scope; then describe the relationship separately in an interpretation note. The relevant relation in the approved training brief is: The second observation complicates attribution; no net treatment effect is provided.

A useful minimality test is to ask whether each standalone sentence remains true if the proposed relation is false. Natural recovery or shared conditions may explain both trajectories. That alternative demonstrates why the words on the page and the link between them require separate auditing. Only the brief for this constructed case resolves the intended link. In an actual translation, identify what language, adjacent document context or author confirmation performs the equivalent role. Do not claim certainty merely because an explanation seems intuitive.

Case 17B. Competing reading and explicitness threshold

Compare the unsafe target, “The treatment caused all the improvement.”, with “Both groups improved; this comparison alone does not establish the intervention’s effect.”. They are not two equally faithful stylistic options under the stated brief. The former introduces a relationship or implication that needs independent justification; the latter retains the intended boundary. A treatment claim requires an appropriate difference and analysis, not just a within-group change. Test the exact meaning of the linking expression, not merely the dictionary meanings of the nouns and verbs. Ask if the target asserts causation, expectation violation, permission, sequence, evidential proof, or exclusivity. If it does, locate the corresponding source commitment.

Where the real document is genuinely ambiguous, there are three defensible editorial routes. Preserve the ambiguity through natural target phrasing, seek a qualified source-side answer, or use an approved translator note that separates explanation from the translated claim. Do not silently invent a causal explanation to make the paragraph feel complete. The specific competing possibility here is: Natural recovery or shared conditions may explain both trajectories. That possibility may or may not be correct, but its continued logical availability shows why an overconfident connective can be dangerous.

Case 17C. Reader-effect and risk diagnostic

A relation error is not just a linguistic blemish; it changes what a reader is licensed to conclude. In this case: Readers may overestimate effectiveness. Review the target as a person relying on the document rather than as an editor looking for elegant wording. Which action, inference, expectation or accusation becomes more likely after the bad version? What would the intended audience believe if the first source proposition were true but the inferred causal or concessive link were false?

A practical severity review should record the precise sentence, the changed inference, the intended reader, and the possible consequence. If the sentence affects safety, rights, access, technical diagnosis, consent or financial decisions, an unsupported connective deserves more scrutiny than an ordinary style improvement. This scenario’s genre is research results, and the source meaning brief is The second observation complicates attribution; no net treatment effect is provided. That is the benchmark for both the accuracy review and the final plain-language edit.

Case 17D. Evidence and author-query gate

An efficient reviewer does not write ‘unclear’ and stop. Ask a question that can be answered with evidence: Are baseline differences and confidence intervals available? Include the source passage and the two candidate target readings. An author can then confirm the relation, provide missing context or explicitly leave the ambiguity open. The record should distinguish an author-confirmed meaning from a translator’s plausible interpretation and from a machine-generated suggestion.

After the answer, update the relation map, not only the visible sentence. If similar segments occur elsewhere, check them for consistency without mechanically propagating the same connective. Context may legitimately make the same pair of vocabulary items express a different relationship in a later document. For this exercise, the brief already supplies the acceptable decision: The second observation complicates attribution; no net treatment effect is provided. The next step is to verify that the chosen target preserves it and creates no new attribution.

Case 17E. Adversarial paraphrase and back-reading

Take the faithful candidate — “Both groups improved; this comparison alone does not establish the intervention’s effect.” — and paraphrase it into a simple reader belief. Now take the unsafe candidate — “The treatment caused all the improvement.” — and do the same. Write down the extra proposition, presupposition or causal claim that appears in the second version. A treatment claim requires an appropriate difference and analysis, not just a within-group change. This comparison is stronger than asking whether both sentences sound natural. A fluent sentence can be misleading, and a back-translation can repeat the same mistake if the reviewer already accepts its assumptions.

For an adversarial check, imagine a reader who accepts the two reported events but rejects the suggested link. Would that reader still be entitled to believe the source? Consider the possibility that natural recovery or shared conditions may explain both trajectories. If yes, the stronger reading needs explicit textual or authorial support. The test does not require mathematical proof. It disciplines translation by making the asserted relationship visible and by preventing an intuition from hiding inside a connective.

Case 17F. Transfer drill and acceptance criterion

Transfer the same reasoning to another language, medium or genre without carrying over English word order. Keep the propositional facts and the stated relation fixed, but choose a natural construction suitable for the target audience. The final acceptable solution does not have to use the English connector in “Both groups improved; this comparison alone does not establish the intervention’s effect.” literally. It does have to preserve the second observation complicates attribution; no net treatment effect is provided. It must also avoid the new belief introduced by “The treatment caused all the improvement.”.

The review is passed when a target reader can identify the two facts, draw only the source-authorized inference, and answer the practical question raised by this research results text without adopting an unsupported explanation. The specific failure warning is: Readers may overestimate effectiveness. Keep the approval record with this sentence’s context, the reviewer’s decision and any author response. If a future revision changes the source relation, rerun this check rather than treating the existing translation as automatically safe.

Case 18: Customer refund offer — Support email

Constructed source meaning: Delivery arrived damaged. The company offered a refund.

Intended meaning supplied by the exercise brief: The offer addresses a reported problem, without necessarily admitting legal liability. Candidate that fails the brief: “The company confessed to negligence and paid damages.” Defensible rendering under the brief: “After the damaged delivery was reported, the company offered a refund.”

Case 18A. Proposition isolation: what is actually stated

In this support email example, begin with the two source propositions without merging them into a single story: Delivery arrived damaged. The company offered a refund. The first analytical task is to preserve each proposition and its evidential status. The passage is not a licence to invent a hidden third event, an extra agent or an unreported decision. A commercial remedy is not automatically a legal admission. A translator who starts by choosing a smooth connective may overlook that the connective itself carries a claim. Start by marking A and B, their subjects and their temporal or logical scope; then describe the relationship separately in an interpretation note. The relevant relation in the approved training brief is: The offer addresses a reported problem, without necessarily admitting legal liability.

A useful minimality test is to ask whether each standalone sentence remains true if the proposed relation is false. The organization may contest fault while still offering a refund. That alternative demonstrates why the words on the page and the link between them require separate auditing. Only the brief for this constructed case resolves the intended link. In an actual translation, identify what language, adjacent document context or author confirmation performs the equivalent role. Do not claim certainty merely because an explanation seems intuitive.

Case 18B. Competing reading and explicitness threshold

Compare the unsafe target, “The company confessed to negligence and paid damages.”, with “After the damaged delivery was reported, the company offered a refund.”. They are not two equally faithful stylistic options under the stated brief. The former introduces a relationship or implication that needs independent justification; the latter retains the intended boundary. A commercial remedy is not automatically a legal admission. Test the exact meaning of the linking expression, not merely the dictionary meanings of the nouns and verbs. Ask if the target asserts causation, expectation violation, permission, sequence, evidential proof, or exclusivity. If it does, locate the corresponding source commitment.

Where the real document is genuinely ambiguous, there are three defensible editorial routes. Preserve the ambiguity through natural target phrasing, seek a qualified source-side answer, or use an approved translator note that separates explanation from the translated claim. Do not silently invent a causal explanation to make the paragraph feel complete. The specific competing possibility here is: The organization may contest fault while still offering a refund. That possibility may or may not be correct, but its continued logical availability shows why an overconfident connective can be dangerous.

Case 18C. Reader-effect and risk diagnostic

A relation error is not just a linguistic blemish; it changes what a reader is licensed to conclude. In this case: Customers may interpret an ordinary remedy as proof of wrongdoing. Review the target as a person relying on the document rather than as an editor looking for elegant wording. Which action, inference, expectation or accusation becomes more likely after the bad version? What would the intended audience believe if the first source proposition were true but the inferred causal or concessive link were false?

A practical severity review should record the precise sentence, the changed inference, the intended reader, and the possible consequence. If the sentence affects safety, rights, access, technical diagnosis, consent or financial decisions, an unsupported connective deserves more scrutiny than an ordinary style improvement. This scenario’s genre is support email, and the source meaning brief is The offer addresses a reported problem, without necessarily admitting legal liability. That is the benchmark for both the accuracy review and the final plain-language edit.

Case 18D. Evidence and author-query gate

An efficient reviewer does not write ‘unclear’ and stop. Ask a question that can be answered with evidence: Is the offer goodwill, contractual entitlement or compensation under a ruling? Include the source passage and the two candidate target readings. An author can then confirm the relation, provide missing context or explicitly leave the ambiguity open. The record should distinguish an author-confirmed meaning from a translator’s plausible interpretation and from a machine-generated suggestion.

After the answer, update the relation map, not only the visible sentence. If similar segments occur elsewhere, check them for consistency without mechanically propagating the same connective. Context may legitimately make the same pair of vocabulary items express a different relationship in a later document. For this exercise, the brief already supplies the acceptable decision: The offer addresses a reported problem, without necessarily admitting legal liability. The next step is to verify that the chosen target preserves it and creates no new attribution.

Case 18E. Adversarial paraphrase and back-reading

Take the faithful candidate — “After the damaged delivery was reported, the company offered a refund.” — and paraphrase it into a simple reader belief. Now take the unsafe candidate — “The company confessed to negligence and paid damages.” — and do the same. Write down the extra proposition, presupposition or causal claim that appears in the second version. A commercial remedy is not automatically a legal admission. This comparison is stronger than asking whether both sentences sound natural. A fluent sentence can be misleading, and a back-translation can repeat the same mistake if the reviewer already accepts its assumptions.

For an adversarial check, imagine a reader who accepts the two reported events but rejects the suggested link. Would that reader still be entitled to believe the source? Consider the possibility that the organization may contest fault while still offering a refund. If yes, the stronger reading needs explicit textual or authorial support. The test does not require mathematical proof. It disciplines translation by making the asserted relationship visible and by preventing an intuition from hiding inside a connective.

Case 18F. Transfer drill and acceptance criterion

Transfer the same reasoning to another language, medium or genre without carrying over English word order. Keep the propositional facts and the stated relation fixed, but choose a natural construction suitable for the target audience. The final acceptable solution does not have to use the English connector in “After the damaged delivery was reported, the company offered a refund.” literally. It does have to preserve the offer addresses a reported problem, without necessarily admitting legal liability. It must also avoid the new belief introduced by “The company confessed to negligence and paid damages.”.

The review is passed when a target reader can identify the two facts, draw only the source-authorized inference, and answer the practical question raised by this support email text without adopting an unsupported explanation. The specific failure warning is: Customers may interpret an ordinary remedy as proof of wrongdoing. Keep the approval record with this sentence’s context, the reviewer’s decision and any author response. If a future revision changes the source relation, rerun this check rather than treating the existing translation as automatically safe.

Case 19: Deadline and acknowledgement — Procedure memo

Constructed source meaning: The request arrived by the deadline. An acknowledgement followed two days later.

Intended meaning supplied by the exercise brief: Arrival complied with the submission deadline; acknowledgement time is separate. Candidate that fails the brief: “The submission was late because acknowledgement took two days.” Defensible rendering under the brief: “The request arrived on time; acknowledgement followed two days later.”

Case 19A. Proposition isolation: what is actually stated

In this procedure memo example, begin with the two source propositions without merging them into a single story: The request arrived by the deadline. An acknowledgement followed two days later. The first analytical task is to preserve each proposition and its evidential status. The passage is not a licence to invent a hidden third event, an extra agent or an unreported decision. The unsafe link mistakes a processing interval for a missed submission obligation. A translator who starts by choosing a smooth connective may overlook that the connective itself carries a claim. Start by marking A and B, their subjects and their temporal or logical scope; then describe the relationship separately in an interpretation note. The relevant relation in the approved training brief is: Arrival complied with the submission deadline; acknowledgement time is separate.

A useful minimality test is to ask whether each standalone sentence remains true if the proposed relation is false. Some programs specify receipt time separately from acceptance time. That alternative demonstrates why the words on the page and the link between them require separate auditing. Only the brief for this constructed case resolves the intended link. In an actual translation, identify what language, adjacent document context or author confirmation performs the equivalent role. Do not claim certainty merely because an explanation seems intuitive.

Case 19B. Competing reading and explicitness threshold

Compare the unsafe target, “The submission was late because acknowledgement took two days.”, with “The request arrived on time; acknowledgement followed two days later.”. They are not two equally faithful stylistic options under the stated brief. The former introduces a relationship or implication that needs independent justification; the latter retains the intended boundary. The unsafe link mistakes a processing interval for a missed submission obligation. Test the exact meaning of the linking expression, not merely the dictionary meanings of the nouns and verbs. Ask if the target asserts causation, expectation violation, permission, sequence, evidential proof, or exclusivity. If it does, locate the corresponding source commitment.

Where the real document is genuinely ambiguous, there are three defensible editorial routes. Preserve the ambiguity through natural target phrasing, seek a qualified source-side answer, or use an approved translator note that separates explanation from the translated claim. Do not silently invent a causal explanation to make the paragraph feel complete. The specific competing possibility here is: Some programs specify receipt time separately from acceptance time. That possibility may or may not be correct, but its continued logical availability shows why an overconfident connective can be dangerous.

Case 19C. Reader-effect and risk diagnostic

A relation error is not just a linguistic blemish; it changes what a reader is licensed to conclude. In this case: A reader could incorrectly lose eligibility. Review the target as a person relying on the document rather than as an editor looking for elegant wording. Which action, inference, expectation or accusation becomes more likely after the bad version? What would the intended audience believe if the first source proposition were true but the inferred causal or concessive link were false?

A practical severity review should record the precise sentence, the changed inference, the intended reader, and the possible consequence. If the sentence affects safety, rights, access, technical diagnosis, consent or financial decisions, an unsupported connective deserves more scrutiny than an ordinary style improvement. This scenario’s genre is procedure memo, and the source meaning brief is Arrival complied with the submission deadline; acknowledgement time is separate. That is the benchmark for both the accuracy review and the final plain-language edit.

Case 19D. Evidence and author-query gate

An efficient reviewer does not write ‘unclear’ and stop. Ask a question that can be answered with evidence: Does acknowledgement date control compliance in this procedure? Include the source passage and the two candidate target readings. An author can then confirm the relation, provide missing context or explicitly leave the ambiguity open. The record should distinguish an author-confirmed meaning from a translator’s plausible interpretation and from a machine-generated suggestion.

After the answer, update the relation map, not only the visible sentence. If similar segments occur elsewhere, check them for consistency without mechanically propagating the same connective. Context may legitimately make the same pair of vocabulary items express a different relationship in a later document. For this exercise, the brief already supplies the acceptable decision: Arrival complied with the submission deadline; acknowledgement time is separate. The next step is to verify that the chosen target preserves it and creates no new attribution.

Case 19E. Adversarial paraphrase and back-reading

Take the faithful candidate — “The request arrived on time; acknowledgement followed two days later.” — and paraphrase it into a simple reader belief. Now take the unsafe candidate — “The submission was late because acknowledgement took two days.” — and do the same. Write down the extra proposition, presupposition or causal claim that appears in the second version. The unsafe link mistakes a processing interval for a missed submission obligation. This comparison is stronger than asking whether both sentences sound natural. A fluent sentence can be misleading, and a back-translation can repeat the same mistake if the reviewer already accepts its assumptions.

For an adversarial check, imagine a reader who accepts the two reported events but rejects the suggested link. Would that reader still be entitled to believe the source? Consider the possibility that some programs specify receipt time separately from acceptance time. If yes, the stronger reading needs explicit textual or authorial support. The test does not require mathematical proof. It disciplines translation by making the asserted relationship visible and by preventing an intuition from hiding inside a connective.

Case 19F. Transfer drill and acceptance criterion

Transfer the same reasoning to another language, medium or genre without carrying over English word order. Keep the propositional facts and the stated relation fixed, but choose a natural construction suitable for the target audience. The final acceptable solution does not have to use the English connector in “The request arrived on time; acknowledgement followed two days later.” literally. It does have to preserve arrival complied with the submission deadline; acknowledgement time is separate. It must also avoid the new belief introduced by “The submission was late because acknowledgement took two days.”.

The review is passed when a target reader can identify the two facts, draw only the source-authorized inference, and answer the practical question raised by this procedure memo text without adopting an unsupported explanation. The specific failure warning is: A reader could incorrectly lose eligibility. Keep the approval record with this sentence’s context, the reviewer’s decision and any author response. If a future revision changes the source relation, rerun this check rather than treating the existing translation as automatically safe.

Case 20: Printer queue after outage — Device status

Constructed source meaning: The printer ran out of paper. The job remained queued.

Intended meaning supplied by the exercise brief: The unresolved queue is attributable to absence of paper under the stated system rule. Candidate that fails the brief: “The printer deleted the job after the paper ran out.” Defensible rendering under the brief: “Because the printer had no paper, the job remained queued.”

Case 20A. Proposition isolation: what is actually stated

In this device status example, begin with the two source propositions without merging them into a single story: The printer ran out of paper. The job remained queued. The first analytical task is to preserve each proposition and its evidential status. The passage is not a licence to invent a hidden third event, an extra agent or an unreported decision. Queued and deleted are different machine states and require different user action. A translator who starts by choosing a smooth connective may overlook that the connective itself carries a claim. Start by marking A and B, their subjects and their temporal or logical scope; then describe the relationship separately in an interpretation note. The relevant relation in the approved training brief is: The unresolved queue is attributable to absence of paper under the stated system rule.

A useful minimality test is to ask whether each standalone sentence remains true if the proposed relation is false. The job could remain queued for reasons other than paper. That alternative demonstrates why the words on the page and the link between them require separate auditing. Only the brief for this constructed case resolves the intended link. In an actual translation, identify what language, adjacent document context or author confirmation performs the equivalent role. Do not claim certainty merely because an explanation seems intuitive.

Case 20B. Competing reading and explicitness threshold

Compare the unsafe target, “The printer deleted the job after the paper ran out.”, with “Because the printer had no paper, the job remained queued.”. They are not two equally faithful stylistic options under the stated brief. The former introduces a relationship or implication that needs independent justification; the latter retains the intended boundary. Queued and deleted are different machine states and require different user action. Test the exact meaning of the linking expression, not merely the dictionary meanings of the nouns and verbs. Ask if the target asserts causation, expectation violation, permission, sequence, evidential proof, or exclusivity. If it does, locate the corresponding source commitment.

Where the real document is genuinely ambiguous, there are three defensible editorial routes. Preserve the ambiguity through natural target phrasing, seek a qualified source-side answer, or use an approved translator note that separates explanation from the translated claim. Do not silently invent a causal explanation to make the paragraph feel complete. The specific competing possibility here is: The job could remain queued for reasons other than paper. That possibility may or may not be correct, but its continued logical availability shows why an overconfident connective can be dangerous.

Case 20C. Reader-effect and risk diagnostic

A relation error is not just a linguistic blemish; it changes what a reader is licensed to conclude. In this case: A user might resend and create duplicates. Review the target as a person relying on the document rather than as an editor looking for elegant wording. Which action, inference, expectation or accusation becomes more likely after the bad version? What would the intended audience believe if the first source proposition were true but the inferred causal or concessive link were false?

A practical severity review should record the precise sentence, the changed inference, the intended reader, and the possible consequence. If the sentence affects safety, rights, access, technical diagnosis, consent or financial decisions, an unsupported connective deserves more scrutiny than an ordinary style improvement. This scenario’s genre is device status, and the source meaning brief is The unresolved queue is attributable to absence of paper under the stated system rule. That is the benchmark for both the accuracy review and the final plain-language edit.

Case 20D. Evidence and author-query gate

An efficient reviewer does not write ‘unclear’ and stop. Ask a question that can be answered with evidence: Does the queue persist through power cycling? Include the source passage and the two candidate target readings. An author can then confirm the relation, provide missing context or explicitly leave the ambiguity open. The record should distinguish an author-confirmed meaning from a translator’s plausible interpretation and from a machine-generated suggestion.

After the answer, update the relation map, not only the visible sentence. If similar segments occur elsewhere, check them for consistency without mechanically propagating the same connective. Context may legitimately make the same pair of vocabulary items express a different relationship in a later document. For this exercise, the brief already supplies the acceptable decision: The unresolved queue is attributable to absence of paper under the stated system rule. The next step is to verify that the chosen target preserves it and creates no new attribution.

Case 20E. Adversarial paraphrase and back-reading

Take the faithful candidate — “Because the printer had no paper, the job remained queued.” — and paraphrase it into a simple reader belief. Now take the unsafe candidate — “The printer deleted the job after the paper ran out.” — and do the same. Write down the extra proposition, presupposition or causal claim that appears in the second version. Queued and deleted are different machine states and require different user action. This comparison is stronger than asking whether both sentences sound natural. A fluent sentence can be misleading, and a back-translation can repeat the same mistake if the reviewer already accepts its assumptions.

For an adversarial check, imagine a reader who accepts the two reported events but rejects the suggested link. Would that reader still be entitled to believe the source? Consider the possibility that the job could remain queued for reasons other than paper. If yes, the stronger reading needs explicit textual or authorial support. The test does not require mathematical proof. It disciplines translation by making the asserted relationship visible and by preventing an intuition from hiding inside a connective.

Case 20F. Transfer drill and acceptance criterion

Transfer the same reasoning to another language, medium or genre without carrying over English word order. Keep the propositional facts and the stated relation fixed, but choose a natural construction suitable for the target audience. The final acceptable solution does not have to use the English connector in “Because the printer had no paper, the job remained queued.” literally. It does have to preserve the unresolved queue is attributable to absence of paper under the stated system rule. It must also avoid the new belief introduced by “The printer deleted the job after the paper ran out.”.

The review is passed when a target reader can identify the two facts, draw only the source-authorized inference, and answer the practical question raised by this device status text without adopting an unsupported explanation. The specific failure warning is: A user might resend and create duplicates. Keep the approval record with this sentence’s context, the reviewer’s decision and any author response. If a future revision changes the source relation, rerun this check rather than treating the existing translation as automatically safe.

Case 21: Forecast and cancelled event — Event announcement

Constructed source meaning: The forecast improved. The event remained cancelled.

Intended meaning supplied by the exercise brief: The statement explicitly contrasts a change in forecast with an unchanged decision. Candidate that fails the brief: “The event was cancelled because the weather improved.” Defensible rendering under the brief: “Although the forecast improved, the event remained cancelled.”

Case 21A. Proposition isolation: what is actually stated

In this event announcement example, begin with the two source propositions without merging them into a single story: The forecast improved. The event remained cancelled. The first analytical task is to preserve each proposition and its evidential status. The passage is not a licence to invent a hidden third event, an extra agent or an unreported decision. The bad bridge reverses an expectation and implies a nonsensical reason. A translator who starts by choosing a smooth connective may overlook that the connective itself carries a claim. Start by marking A and B, their subjects and their temporal or logical scope; then describe the relationship separately in an interpretation note. The relevant relation in the approved training brief is: The statement explicitly contrasts a change in forecast with an unchanged decision.

A useful minimality test is to ask whether each standalone sentence remains true if the proposed relation is false. The organizer may have other constraints. That alternative demonstrates why the words on the page and the link between them require separate auditing. Only the brief for this constructed case resolves the intended link. In an actual translation, identify what language, adjacent document context or author confirmation performs the equivalent role. Do not claim certainty merely because an explanation seems intuitive.

Case 21B. Competing reading and explicitness threshold

Compare the unsafe target, “The event was cancelled because the weather improved.”, with “Although the forecast improved, the event remained cancelled.”. They are not two equally faithful stylistic options under the stated brief. The former introduces a relationship or implication that needs independent justification; the latter retains the intended boundary. The bad bridge reverses an expectation and implies a nonsensical reason. Test the exact meaning of the linking expression, not merely the dictionary meanings of the nouns and verbs. Ask if the target asserts causation, expectation violation, permission, sequence, evidential proof, or exclusivity. If it does, locate the corresponding source commitment.

Where the real document is genuinely ambiguous, there are three defensible editorial routes. Preserve the ambiguity through natural target phrasing, seek a qualified source-side answer, or use an approved translator note that separates explanation from the translated claim. Do not silently invent a causal explanation to make the paragraph feel complete. The specific competing possibility here is: The organizer may have other constraints. That possibility may or may not be correct, but its continued logical availability shows why an overconfident connective can be dangerous.

Case 21C. Reader-effect and risk diagnostic

A relation error is not just a linguistic blemish; it changes what a reader is licensed to conclude. In this case: Guests may think the cancellation was a penalty for good weather. Review the target as a person relying on the document rather than as an editor looking for elegant wording. Which action, inference, expectation or accusation becomes more likely after the bad version? What would the intended audience believe if the first source proposition were true but the inferred causal or concessive link were false?

A practical severity review should record the precise sentence, the changed inference, the intended reader, and the possible consequence. If the sentence affects safety, rights, access, technical diagnosis, consent or financial decisions, an unsupported connective deserves more scrutiny than an ordinary style improvement. This scenario’s genre is event announcement, and the source meaning brief is The statement explicitly contrasts a change in forecast with an unchanged decision. That is the benchmark for both the accuracy review and the final plain-language edit.

Case 21D. Evidence and author-query gate

An efficient reviewer does not write ‘unclear’ and stop. Ask a question that can be answered with evidence: Was cancellation irreversible due to contractual or logistical factors? Include the source passage and the two candidate target readings. An author can then confirm the relation, provide missing context or explicitly leave the ambiguity open. The record should distinguish an author-confirmed meaning from a translator’s plausible interpretation and from a machine-generated suggestion.

After the answer, update the relation map, not only the visible sentence. If similar segments occur elsewhere, check them for consistency without mechanically propagating the same connective. Context may legitimately make the same pair of vocabulary items express a different relationship in a later document. For this exercise, the brief already supplies the acceptable decision: The statement explicitly contrasts a change in forecast with an unchanged decision. The next step is to verify that the chosen target preserves it and creates no new attribution.

Case 21E. Adversarial paraphrase and back-reading

Take the faithful candidate — “Although the forecast improved, the event remained cancelled.” — and paraphrase it into a simple reader belief. Now take the unsafe candidate — “The event was cancelled because the weather improved.” — and do the same. Write down the extra proposition, presupposition or causal claim that appears in the second version. The bad bridge reverses an expectation and implies a nonsensical reason. This comparison is stronger than asking whether both sentences sound natural. A fluent sentence can be misleading, and a back-translation can repeat the same mistake if the reviewer already accepts its assumptions.

For an adversarial check, imagine a reader who accepts the two reported events but rejects the suggested link. Would that reader still be entitled to believe the source? Consider the possibility that the organizer may have other constraints. If yes, the stronger reading needs explicit textual or authorial support. The test does not require mathematical proof. It disciplines translation by making the asserted relationship visible and by preventing an intuition from hiding inside a connective.

Case 21F. Transfer drill and acceptance criterion

Transfer the same reasoning to another language, medium or genre without carrying over English word order. Keep the propositional facts and the stated relation fixed, but choose a natural construction suitable for the target audience. The final acceptable solution does not have to use the English connector in “Although the forecast improved, the event remained cancelled.” literally. It does have to preserve the statement explicitly contrasts a change in forecast with an unchanged decision. It must also avoid the new belief introduced by “The event was cancelled because the weather improved.”.

The review is passed when a target reader can identify the two facts, draw only the source-authorized inference, and answer the practical question raised by this event announcement text without adopting an unsupported explanation. The specific failure warning is: Guests may think the cancellation was a penalty for good weather. Keep the approval record with this sentence’s context, the reviewer’s decision and any author response. If a future revision changes the source relation, rerun this check rather than treating the existing translation as automatically safe.

Case 22: An abstract and its example — Learning guide

Constructed source meaning: Several factors influence comprehension. Prior vocabulary is one factor.

Intended meaning supplied by the exercise brief: The second sentence gives an instance rather than the only factor. Candidate that fails the brief: “Comprehension depends solely on prior vocabulary.” Defensible rendering under the brief: “Several factors influence comprehension; prior vocabulary is one example.”

Case 22A. Proposition isolation: what is actually stated

In this learning guide example, begin with the two source propositions without merging them into a single story: Several factors influence comprehension. Prior vocabulary is one factor. The first analytical task is to preserve each proposition and its evidential status. The passage is not a licence to invent a hidden third event, an extra agent or an unreported decision. An illustration must not become an exhaustive definition. A translator who starts by choosing a smooth connective may overlook that the connective itself carries a claim. Start by marking A and B, their subjects and their temporal or logical scope; then describe the relationship separately in an interpretation note. The relevant relation in the approved training brief is: The second sentence gives an instance rather than the only factor.

A useful minimality test is to ask whether each standalone sentence remains true if the proposed relation is false. Other listed factors could have different relative importance. That alternative demonstrates why the words on the page and the link between them require separate auditing. Only the brief for this constructed case resolves the intended link. In an actual translation, identify what language, adjacent document context or author confirmation performs the equivalent role. Do not claim certainty merely because an explanation seems intuitive.

Case 22B. Competing reading and explicitness threshold

Compare the unsafe target, “Comprehension depends solely on prior vocabulary.”, with “Several factors influence comprehension; prior vocabulary is one example.”. They are not two equally faithful stylistic options under the stated brief. The former introduces a relationship or implication that needs independent justification; the latter retains the intended boundary. An illustration must not become an exhaustive definition. Test the exact meaning of the linking expression, not merely the dictionary meanings of the nouns and verbs. Ask if the target asserts causation, expectation violation, permission, sequence, evidential proof, or exclusivity. If it does, locate the corresponding source commitment.

Where the real document is genuinely ambiguous, there are three defensible editorial routes. Preserve the ambiguity through natural target phrasing, seek a qualified source-side answer, or use an approved translator note that separates explanation from the translated claim. Do not silently invent a causal explanation to make the paragraph feel complete. The specific competing possibility here is: Other listed factors could have different relative importance. That possibility may or may not be correct, but its continued logical availability shows why an overconfident connective can be dangerous.

Case 22C. Reader-effect and risk diagnostic

A relation error is not just a linguistic blemish; it changes what a reader is licensed to conclude. In this case: Readers might ignore attention, background knowledge or text structure. Review the target as a person relying on the document rather than as an editor looking for elegant wording. Which action, inference, expectation or accusation becomes more likely after the bad version? What would the intended audience believe if the first source proposition were true but the inferred causal or concessive link were false?

A practical severity review should record the precise sentence, the changed inference, the intended reader, and the possible consequence. If the sentence affects safety, rights, access, technical diagnosis, consent or financial decisions, an unsupported connective deserves more scrutiny than an ordinary style improvement. This scenario’s genre is learning guide, and the source meaning brief is The second sentence gives an instance rather than the only factor. That is the benchmark for both the accuracy review and the final plain-language edit.

Case 22D. Evidence and author-query gate

An efficient reviewer does not write ‘unclear’ and stop. Ask a question that can be answered with evidence: Is vocabulary the chief example or one among many? Include the source passage and the two candidate target readings. An author can then confirm the relation, provide missing context or explicitly leave the ambiguity open. The record should distinguish an author-confirmed meaning from a translator’s plausible interpretation and from a machine-generated suggestion.

After the answer, update the relation map, not only the visible sentence. If similar segments occur elsewhere, check them for consistency without mechanically propagating the same connective. Context may legitimately make the same pair of vocabulary items express a different relationship in a later document. For this exercise, the brief already supplies the acceptable decision: The second sentence gives an instance rather than the only factor. The next step is to verify that the chosen target preserves it and creates no new attribution.

Case 22E. Adversarial paraphrase and back-reading

Take the faithful candidate — “Several factors influence comprehension; prior vocabulary is one example.” — and paraphrase it into a simple reader belief. Now take the unsafe candidate — “Comprehension depends solely on prior vocabulary.” — and do the same. Write down the extra proposition, presupposition or causal claim that appears in the second version. An illustration must not become an exhaustive definition. This comparison is stronger than asking whether both sentences sound natural. A fluent sentence can be misleading, and a back-translation can repeat the same mistake if the reviewer already accepts its assumptions.

For an adversarial check, imagine a reader who accepts the two reported events but rejects the suggested link. Would that reader still be entitled to believe the source? Consider the possibility that other listed factors could have different relative importance. If yes, the stronger reading needs explicit textual or authorial support. The test does not require mathematical proof. It disciplines translation by making the asserted relationship visible and by preventing an intuition from hiding inside a connective.

Case 22F. Transfer drill and acceptance criterion

Transfer the same reasoning to another language, medium or genre without carrying over English word order. Keep the propositional facts and the stated relation fixed, but choose a natural construction suitable for the target audience. The final acceptable solution does not have to use the English connector in “Several factors influence comprehension; prior vocabulary is one example.” literally. It does have to preserve the second sentence gives an instance rather than the only factor. It must also avoid the new belief introduced by “Comprehension depends solely on prior vocabulary.”.

The review is passed when a target reader can identify the two facts, draw only the source-authorized inference, and answer the practical question raised by this learning guide text without adopting an unsupported explanation. The specific failure warning is: Readers might ignore attention, background knowledge or text structure. Keep the approval record with this sentence’s context, the reviewer’s decision and any author response. If a future revision changes the source relation, rerun this check rather than treating the existing translation as automatically safe.

Case 23: A patient instructions label — Health information

Constructed source meaning: The leaflet says to use the medicine with food. A pharmacist offers further instructions.

Intended meaning supplied by the exercise brief: The second sentence is an additional support route, not a claim that pharmacists substitute for printed instructions. Candidate that fails the brief: “Ignore the leaflet if a pharmacist is available.” Defensible rendering under the brief: “Follow the leaflet’s instructions and ask a pharmacist for clarification if needed.”

Case 23A. Proposition isolation: what is actually stated

In this health information example, begin with the two source propositions without merging them into a single story: The leaflet says to use the medicine with food. A pharmacist offers further instructions. The first analytical task is to preserve each proposition and its evidential status. The passage is not a licence to invent a hidden third event, an extra agent or an unreported decision. Changing coordination into an alternative could produce unsafe reader behavior. A translator who starts by choosing a smooth connective may overlook that the connective itself carries a claim. Start by marking A and B, their subjects and their temporal or logical scope; then describe the relationship separately in an interpretation note. The relevant relation in the approved training brief is: The second sentence is an additional support route, not a claim that pharmacists substitute for printed instructions.

A useful minimality test is to ask whether each standalone sentence remains true if the proposed relation is false. Different medicines have different requirements; this is only a discourse example. That alternative demonstrates why the words on the page and the link between them require separate auditing. Only the brief for this constructed case resolves the intended link. In an actual translation, identify what language, adjacent document context or author confirmation performs the equivalent role. Do not claim certainty merely because an explanation seems intuitive.

Case 23B. Competing reading and explicitness threshold

Compare the unsafe target, “Ignore the leaflet if a pharmacist is available.”, with “Follow the leaflet’s instructions and ask a pharmacist for clarification if needed.”. They are not two equally faithful stylistic options under the stated brief. The former introduces a relationship or implication that needs independent justification; the latter retains the intended boundary. Changing coordination into an alternative could produce unsafe reader behavior. Test the exact meaning of the linking expression, not merely the dictionary meanings of the nouns and verbs. Ask if the target asserts causation, expectation violation, permission, sequence, evidential proof, or exclusivity. If it does, locate the corresponding source commitment.

Where the real document is genuinely ambiguous, there are three defensible editorial routes. Preserve the ambiguity through natural target phrasing, seek a qualified source-side answer, or use an approved translator note that separates explanation from the translated claim. Do not silently invent a causal explanation to make the paragraph feel complete. The specific competing possibility here is: Different medicines have different requirements; this is only a discourse example. That possibility may or may not be correct, but its continued logical availability shows why an overconfident connective can be dangerous.

Case 23C. Reader-effect and risk diagnostic

A relation error is not just a linguistic blemish; it changes what a reader is licensed to conclude. In this case: Readers may skip essential dosing instructions. Review the target as a person relying on the document rather than as an editor looking for elegant wording. Which action, inference, expectation or accusation becomes more likely after the bad version? What would the intended audience believe if the first source proposition were true but the inferred causal or concessive link were false?

A practical severity review should record the precise sentence, the changed inference, the intended reader, and the possible consequence. If the sentence affects safety, rights, access, technical diagnosis, consent or financial decisions, an unsupported connective deserves more scrutiny than an ordinary style improvement. This scenario’s genre is health information, and the source meaning brief is The second sentence is an additional support route, not a claim that pharmacists substitute for printed instructions. That is the benchmark for both the accuracy review and the final plain-language edit.

Case 23D. Evidence and author-query gate

An efficient reviewer does not write ‘unclear’ and stop. Ask a question that can be answered with evidence: Does the leaflet itself give an exception or conflict-resolution process? Include the source passage and the two candidate target readings. An author can then confirm the relation, provide missing context or explicitly leave the ambiguity open. The record should distinguish an author-confirmed meaning from a translator’s plausible interpretation and from a machine-generated suggestion.

After the answer, update the relation map, not only the visible sentence. If similar segments occur elsewhere, check them for consistency without mechanically propagating the same connective. Context may legitimately make the same pair of vocabulary items express a different relationship in a later document. For this exercise, the brief already supplies the acceptable decision: The second sentence is an additional support route, not a claim that pharmacists substitute for printed instructions. The next step is to verify that the chosen target preserves it and creates no new attribution.

Case 23E. Adversarial paraphrase and back-reading

Take the faithful candidate — “Follow the leaflet’s instructions and ask a pharmacist for clarification if needed.” — and paraphrase it into a simple reader belief. Now take the unsafe candidate — “Ignore the leaflet if a pharmacist is available.” — and do the same. Write down the extra proposition, presupposition or causal claim that appears in the second version. Changing coordination into an alternative could produce unsafe reader behavior. This comparison is stronger than asking whether both sentences sound natural. A fluent sentence can be misleading, and a back-translation can repeat the same mistake if the reviewer already accepts its assumptions.

For an adversarial check, imagine a reader who accepts the two reported events but rejects the suggested link. Would that reader still be entitled to believe the source? Consider the possibility that different medicines have different requirements; this is only a discourse example. If yes, the stronger reading needs explicit textual or authorial support. The test does not require mathematical proof. It disciplines translation by making the asserted relationship visible and by preventing an intuition from hiding inside a connective.

Case 23F. Transfer drill and acceptance criterion

Transfer the same reasoning to another language, medium or genre without carrying over English word order. Keep the propositional facts and the stated relation fixed, but choose a natural construction suitable for the target audience. The final acceptable solution does not have to use the English connector in “Follow the leaflet’s instructions and ask a pharmacist for clarification if needed.” literally. It does have to preserve the second sentence is an additional support route, not a claim that pharmacists substitute for printed instructions. It must also avoid the new belief introduced by “Ignore the leaflet if a pharmacist is available.”.

The review is passed when a target reader can identify the two facts, draw only the source-authorized inference, and answer the practical question raised by this health information text without adopting an unsupported explanation. The specific failure warning is: Readers may skip essential dosing instructions. Keep the approval record with this sentence’s context, the reviewer’s decision and any author response. If a future revision changes the source relation, rerun this check rather than treating the existing translation as automatically safe.

Case 24: Administrative permission — Policy summary

Constructed source meaning: Members may request a review. The committee decides whether to grant one.

Intended meaning supplied by the exercise brief: The first clause grants permission to apply, not an entitlement to approval. Candidate that fails the brief: “Every review request must be granted by the committee.” Defensible rendering under the brief: “Members may request a review; the committee retains discretion over approval.”

Case 24A. Proposition isolation: what is actually stated

In this policy summary example, begin with the two source propositions without merging them into a single story: Members may request a review. The committee decides whether to grant one. The first analytical task is to preserve each proposition and its evidential status. The passage is not a licence to invent a hidden third event, an extra agent or an unreported decision. Opportunity to request is logically weaker than right to obtain. A translator who starts by choosing a smooth connective may overlook that the connective itself carries a claim. Start by marking A and B, their subjects and their temporal or logical scope; then describe the relationship separately in an interpretation note. The relevant relation in the approved training brief is: The first clause grants permission to apply, not an entitlement to approval.

A useful minimality test is to ask whether each standalone sentence remains true if the proposed relation is false. A separate regulation may constrain discretion. That alternative demonstrates why the words on the page and the link between them require separate auditing. Only the brief for this constructed case resolves the intended link. In an actual translation, identify what language, adjacent document context or author confirmation performs the equivalent role. Do not claim certainty merely because an explanation seems intuitive.

Case 24B. Competing reading and explicitness threshold

Compare the unsafe target, “Every review request must be granted by the committee.”, with “Members may request a review; the committee retains discretion over approval.”. They are not two equally faithful stylistic options under the stated brief. The former introduces a relationship or implication that needs independent justification; the latter retains the intended boundary. Opportunity to request is logically weaker than right to obtain. Test the exact meaning of the linking expression, not merely the dictionary meanings of the nouns and verbs. Ask if the target asserts causation, expectation violation, permission, sequence, evidential proof, or exclusivity. If it does, locate the corresponding source commitment.

Where the real document is genuinely ambiguous, there are three defensible editorial routes. Preserve the ambiguity through natural target phrasing, seek a qualified source-side answer, or use an approved translator note that separates explanation from the translated claim. Do not silently invent a causal explanation to make the paragraph feel complete. The specific competing possibility here is: A separate regulation may constrain discretion. That possibility may or may not be correct, but its continued logical availability shows why an overconfident connective can be dangerous.

Case 24C. Reader-effect and risk diagnostic

A relation error is not just a linguistic blemish; it changes what a reader is licensed to conclude. In this case: Readers may form incorrect expectations about outcomes. Review the target as a person relying on the document rather than as an editor looking for elegant wording. Which action, inference, expectation or accusation becomes more likely after the bad version? What would the intended audience believe if the first source proposition were true but the inferred causal or concessive link were false?

A practical severity review should record the precise sentence, the changed inference, the intended reader, and the possible consequence. If the sentence affects safety, rights, access, technical diagnosis, consent or financial decisions, an unsupported connective deserves more scrutiny than an ordinary style improvement. This scenario’s genre is policy summary, and the source meaning brief is The first clause grants permission to apply, not an entitlement to approval. That is the benchmark for both the accuracy review and the final plain-language edit.

Case 24D. Evidence and author-query gate

An efficient reviewer does not write ‘unclear’ and stop. Ask a question that can be answered with evidence: Does any clause prescribe approval criteria or appeal rights? Include the source passage and the two candidate target readings. An author can then confirm the relation, provide missing context or explicitly leave the ambiguity open. The record should distinguish an author-confirmed meaning from a translator’s plausible interpretation and from a machine-generated suggestion.

After the answer, update the relation map, not only the visible sentence. If similar segments occur elsewhere, check them for consistency without mechanically propagating the same connective. Context may legitimately make the same pair of vocabulary items express a different relationship in a later document. For this exercise, the brief already supplies the acceptable decision: The first clause grants permission to apply, not an entitlement to approval. The next step is to verify that the chosen target preserves it and creates no new attribution.

Case 24E. Adversarial paraphrase and back-reading

Take the faithful candidate — “Members may request a review; the committee retains discretion over approval.” — and paraphrase it into a simple reader belief. Now take the unsafe candidate — “Every review request must be granted by the committee.” — and do the same. Write down the extra proposition, presupposition or causal claim that appears in the second version. Opportunity to request is logically weaker than right to obtain. This comparison is stronger than asking whether both sentences sound natural. A fluent sentence can be misleading, and a back-translation can repeat the same mistake if the reviewer already accepts its assumptions.

For an adversarial check, imagine a reader who accepts the two reported events but rejects the suggested link. Would that reader still be entitled to believe the source? Consider the possibility that a separate regulation may constrain discretion. If yes, the stronger reading needs explicit textual or authorial support. The test does not require mathematical proof. It disciplines translation by making the asserted relationship visible and by preventing an intuition from hiding inside a connective.

Case 24F. Transfer drill and acceptance criterion

Transfer the same reasoning to another language, medium or genre without carrying over English word order. Keep the propositional facts and the stated relation fixed, but choose a natural construction suitable for the target audience. The final acceptable solution does not have to use the English connector in “Members may request a review; the committee retains discretion over approval.” literally. It does have to preserve the first clause grants permission to apply, not an entitlement to approval. It must also avoid the new belief introduced by “Every review request must be granted by the committee.”.

The review is passed when a target reader can identify the two facts, draw only the source-authorized inference, and answer the practical question raised by this policy summary text without adopting an unsupported explanation. The specific failure warning is: Readers may form incorrect expectations about outcomes. Keep the approval record with this sentence’s context, the reviewer’s decision and any author response. If a future revision changes the source relation, rerun this check rather than treating the existing translation as automatically safe.

Case 25: Archived record correction — Historical note

Constructed source meaning: A newly found record gives a different date. The exhibition label was revised.

Intended meaning supplied by the exercise brief: The revision responded to new evidence; the source does not claim the old label was fabricated. Candidate that fails the brief: “The original label was intentionally false and had to be removed.” Defensible rendering under the brief: “After the new record was assessed, the exhibition label was revised.”

Case 25A. Proposition isolation: what is actually stated

In this historical note example, begin with the two source propositions without merging them into a single story: A newly found record gives a different date. The exhibition label was revised. The first analytical task is to preserve each proposition and its evidential status. The passage is not a licence to invent a hidden third event, an extra agent or an unreported decision. Correction of information does not imply deliberate deception. A translator who starts by choosing a smooth connective may overlook that the connective itself carries a claim. Start by marking A and B, their subjects and their temporal or logical scope; then describe the relationship separately in an interpretation note. The relevant relation in the approved training brief is: The revision responded to new evidence; the source does not claim the old label was fabricated.

A useful minimality test is to ask whether each standalone sentence remains true if the proposed relation is false. The previous date may have been a reasonable historical estimate. That alternative demonstrates why the words on the page and the link between them require separate auditing. Only the brief for this constructed case resolves the intended link. In an actual translation, identify what language, adjacent document context or author confirmation performs the equivalent role. Do not claim certainty merely because an explanation seems intuitive.

Case 25B. Competing reading and explicitness threshold

Compare the unsafe target, “The original label was intentionally false and had to be removed.”, with “After the new record was assessed, the exhibition label was revised.”. They are not two equally faithful stylistic options under the stated brief. The former introduces a relationship or implication that needs independent justification; the latter retains the intended boundary. Correction of information does not imply deliberate deception. Test the exact meaning of the linking expression, not merely the dictionary meanings of the nouns and verbs. Ask if the target asserts causation, expectation violation, permission, sequence, evidential proof, or exclusivity. If it does, locate the corresponding source commitment.

Where the real document is genuinely ambiguous, there are three defensible editorial routes. Preserve the ambiguity through natural target phrasing, seek a qualified source-side answer, or use an approved translator note that separates explanation from the translated claim. Do not silently invent a causal explanation to make the paragraph feel complete. The specific competing possibility here is: The previous date may have been a reasonable historical estimate. That possibility may or may not be correct, but its continued logical availability shows why an overconfident connective can be dangerous.

Case 25C. Reader-effect and risk diagnostic

A relation error is not just a linguistic blemish; it changes what a reader is licensed to conclude. In this case: Readers may accuse curators of misconduct without evidence. Review the target as a person relying on the document rather than as an editor looking for elegant wording. Which action, inference, expectation or accusation becomes more likely after the bad version? What would the intended audience believe if the first source proposition were true but the inferred causal or concessive link were false?

A practical severity review should record the precise sentence, the changed inference, the intended reader, and the possible consequence. If the sentence affects safety, rights, access, technical diagnosis, consent or financial decisions, an unsupported connective deserves more scrutiny than an ordinary style improvement. This scenario’s genre is historical note, and the source meaning brief is The revision responded to new evidence; the source does not claim the old label was fabricated. That is the benchmark for both the accuracy review and the final plain-language edit.

Case 25D. Evidence and author-query gate

An efficient reviewer does not write ‘unclear’ and stop. Ask a question that can be answered with evidence: Was the new record decisive or one of several sources? Include the source passage and the two candidate target readings. An author can then confirm the relation, provide missing context or explicitly leave the ambiguity open. The record should distinguish an author-confirmed meaning from a translator’s plausible interpretation and from a machine-generated suggestion.

After the answer, update the relation map, not only the visible sentence. If similar segments occur elsewhere, check them for consistency without mechanically propagating the same connective. Context may legitimately make the same pair of vocabulary items express a different relationship in a later document. For this exercise, the brief already supplies the acceptable decision: The revision responded to new evidence; the source does not claim the old label was fabricated. The next step is to verify that the chosen target preserves it and creates no new attribution.

Case 25E. Adversarial paraphrase and back-reading

Take the faithful candidate — “After the new record was assessed, the exhibition label was revised.” — and paraphrase it into a simple reader belief. Now take the unsafe candidate — “The original label was intentionally false and had to be removed.” — and do the same. Write down the extra proposition, presupposition or causal claim that appears in the second version. Correction of information does not imply deliberate deception. This comparison is stronger than asking whether both sentences sound natural. A fluent sentence can be misleading, and a back-translation can repeat the same mistake if the reviewer already accepts its assumptions.

For an adversarial check, imagine a reader who accepts the two reported events but rejects the suggested link. Would that reader still be entitled to believe the source? Consider the possibility that the previous date may have been a reasonable historical estimate. If yes, the stronger reading needs explicit textual or authorial support. The test does not require mathematical proof. It disciplines translation by making the asserted relationship visible and by preventing an intuition from hiding inside a connective.

Case 25F. Transfer drill and acceptance criterion

Transfer the same reasoning to another language, medium or genre without carrying over English word order. Keep the propositional facts and the stated relation fixed, but choose a natural construction suitable for the target audience. The final acceptable solution does not have to use the English connector in “After the new record was assessed, the exhibition label was revised.” literally. It does have to preserve the revision responded to new evidence; the source does not claim the old label was fabricated. It must also avoid the new belief introduced by “The original label was intentionally false and had to be removed.”.

The review is passed when a target reader can identify the two facts, draw only the source-authorized inference, and answer the practical question raised by this historical note text without adopting an unsupported explanation. The specific failure warning is: Readers may accuse curators of misconduct without evidence. Keep the approval record with this sentence’s context, the reviewer’s decision and any author response. If a future revision changes the source relation, rerun this check rather than treating the existing translation as automatically safe.

Case 26: Supplier price and contract — Business report

Constructed source meaning: The supplier raised its list price. The existing contract stayed unchanged.

Intended meaning supplied by the exercise brief: List pricing and contractual pricing are distinct; the second fact limits what changed. Candidate that fails the brief: “The contract automatically became more expensive.” Defensible rendering under the brief: “The supplier raised its list price, while the existing contract remained unchanged.”

Case 26A. Proposition isolation: what is actually stated

In this business report example, begin with the two source propositions without merging them into a single story: The supplier raised its list price. The existing contract stayed unchanged. The first analytical task is to preserve each proposition and its evidential status. The passage is not a licence to invent a hidden third event, an extra agent or an unreported decision. A change in general price is not automatically a contract amendment. A translator who starts by choosing a smooth connective may overlook that the connective itself carries a claim. Start by marking A and B, their subjects and their temporal or logical scope; then describe the relationship separately in an interpretation note. The relevant relation in the approved training brief is: List pricing and contractual pricing are distinct; the second fact limits what changed.

A useful minimality test is to ask whether each standalone sentence remains true if the proposed relation is false. A renewal could later use the new price. That alternative demonstrates why the words on the page and the link between them require separate auditing. Only the brief for this constructed case resolves the intended link. In an actual translation, identify what language, adjacent document context or author confirmation performs the equivalent role. Do not claim certainty merely because an explanation seems intuitive.

Case 26B. Competing reading and explicitness threshold

Compare the unsafe target, “The contract automatically became more expensive.”, with “The supplier raised its list price, while the existing contract remained unchanged.”. They are not two equally faithful stylistic options under the stated brief. The former introduces a relationship or implication that needs independent justification; the latter retains the intended boundary. A change in general price is not automatically a contract amendment. Test the exact meaning of the linking expression, not merely the dictionary meanings of the nouns and verbs. Ask if the target asserts causation, expectation violation, permission, sequence, evidential proof, or exclusivity. If it does, locate the corresponding source commitment.

Where the real document is genuinely ambiguous, there are three defensible editorial routes. Preserve the ambiguity through natural target phrasing, seek a qualified source-side answer, or use an approved translator note that separates explanation from the translated claim. Do not silently invent a causal explanation to make the paragraph feel complete. The specific competing possibility here is: A renewal could later use the new price. That possibility may or may not be correct, but its continued logical availability shows why an overconfident connective can be dangerous.

Case 26C. Reader-effect and risk diagnostic

A relation error is not just a linguistic blemish; it changes what a reader is licensed to conclude. In this case: Buyers could overstate costs or payment obligations. Review the target as a person relying on the document rather than as an editor looking for elegant wording. Which action, inference, expectation or accusation becomes more likely after the bad version? What would the intended audience believe if the first source proposition were true but the inferred causal or concessive link were false?

A practical severity review should record the precise sentence, the changed inference, the intended reader, and the possible consequence. If the sentence affects safety, rights, access, technical diagnosis, consent or financial decisions, an unsupported connective deserves more scrutiny than an ordinary style improvement. This scenario’s genre is business report, and the source meaning brief is List pricing and contractual pricing are distinct; the second fact limits what changed. That is the benchmark for both the accuracy review and the final plain-language edit.

Case 26D. Evidence and author-query gate

An efficient reviewer does not write ‘unclear’ and stop. Ask a question that can be answered with evidence: Is there an escalation clause in the signed contract? Include the source passage and the two candidate target readings. An author can then confirm the relation, provide missing context or explicitly leave the ambiguity open. The record should distinguish an author-confirmed meaning from a translator’s plausible interpretation and from a machine-generated suggestion.

After the answer, update the relation map, not only the visible sentence. If similar segments occur elsewhere, check them for consistency without mechanically propagating the same connective. Context may legitimately make the same pair of vocabulary items express a different relationship in a later document. For this exercise, the brief already supplies the acceptable decision: List pricing and contractual pricing are distinct; the second fact limits what changed. The next step is to verify that the chosen target preserves it and creates no new attribution.

Case 26E. Adversarial paraphrase and back-reading

Take the faithful candidate — “The supplier raised its list price, while the existing contract remained unchanged.” — and paraphrase it into a simple reader belief. Now take the unsafe candidate — “The contract automatically became more expensive.” — and do the same. Write down the extra proposition, presupposition or causal claim that appears in the second version. A change in general price is not automatically a contract amendment. This comparison is stronger than asking whether both sentences sound natural. A fluent sentence can be misleading, and a back-translation can repeat the same mistake if the reviewer already accepts its assumptions.

For an adversarial check, imagine a reader who accepts the two reported events but rejects the suggested link. Would that reader still be entitled to believe the source? Consider the possibility that a renewal could later use the new price. If yes, the stronger reading needs explicit textual or authorial support. The test does not require mathematical proof. It disciplines translation by making the asserted relationship visible and by preventing an intuition from hiding inside a connective.

Case 26F. Transfer drill and acceptance criterion

Transfer the same reasoning to another language, medium or genre without carrying over English word order. Keep the propositional facts and the stated relation fixed, but choose a natural construction suitable for the target audience. The final acceptable solution does not have to use the English connector in “The supplier raised its list price, while the existing contract remained unchanged.” literally. It does have to preserve list pricing and contractual pricing are distinct; the second fact limits what changed. It must also avoid the new belief introduced by “The contract automatically became more expensive.”.

The review is passed when a target reader can identify the two facts, draw only the source-authorized inference, and answer the practical question raised by this business report text without adopting an unsupported explanation. The specific failure warning is: Buyers could overstate costs or payment obligations. Keep the approval record with this sentence’s context, the reviewer’s decision and any author response. If a future revision changes the source relation, rerun this check rather than treating the existing translation as automatically safe.

Case 27: Route opening and arrival time — Operations log

Constructed source meaning: A new track section opened. Typical arrival times stayed similar.

Intended meaning supplied by the exercise brief: A potentially expected improvement did not appear; causal efficacy is not claimed. Candidate that fails the brief: “The new track made trains arrive slower.” Defensible rendering under the brief: “Despite the new track section opening, typical arrival times remained similar.”

Case 27A. Proposition isolation: what is actually stated

In this operations log example, begin with the two source propositions without merging them into a single story: A new track section opened. Typical arrival times stayed similar. The first analytical task is to preserve each proposition and its evidential status. The passage is not a licence to invent a hidden third event, an extra agent or an unreported decision. The target should not invent deterioration from the absence of improvement. A translator who starts by choosing a smooth connective may overlook that the connective itself carries a claim. Start by marking A and B, their subjects and their temporal or logical scope; then describe the relationship separately in an interpretation note. The relevant relation in the approved training brief is: A potentially expected improvement did not appear; causal efficacy is not claimed.

A useful minimality test is to ask whether each standalone sentence remains true if the proposed relation is false. Opening capacity may affect resilience rather than speed. That alternative demonstrates why the words on the page and the link between them require separate auditing. Only the brief for this constructed case resolves the intended link. In an actual translation, identify what language, adjacent document context or author confirmation performs the equivalent role. Do not claim certainty merely because an explanation seems intuitive.

Case 27B. Competing reading and explicitness threshold

Compare the unsafe target, “The new track made trains arrive slower.”, with “Despite the new track section opening, typical arrival times remained similar.”. They are not two equally faithful stylistic options under the stated brief. The former introduces a relationship or implication that needs independent justification; the latter retains the intended boundary. The target should not invent deterioration from the absence of improvement. Test the exact meaning of the linking expression, not merely the dictionary meanings of the nouns and verbs. Ask if the target asserts causation, expectation violation, permission, sequence, evidential proof, or exclusivity. If it does, locate the corresponding source commitment.

Where the real document is genuinely ambiguous, there are three defensible editorial routes. Preserve the ambiguity through natural target phrasing, seek a qualified source-side answer, or use an approved translator note that separates explanation from the translated claim. Do not silently invent a causal explanation to make the paragraph feel complete. The specific competing possibility here is: Opening capacity may affect resilience rather than speed. That possibility may or may not be correct, but its continued logical availability shows why an overconfident connective can be dangerous.

Case 27C. Reader-effect and risk diagnostic

A relation error is not just a linguistic blemish; it changes what a reader is licensed to conclude. In this case: Decision makers might misjudge an investment. Review the target as a person relying on the document rather than as an editor looking for elegant wording. Which action, inference, expectation or accusation becomes more likely after the bad version? What would the intended audience believe if the first source proposition were true but the inferred causal or concessive link were false?

A practical severity review should record the precise sentence, the changed inference, the intended reader, and the possible consequence. If the sentence affects safety, rights, access, technical diagnosis, consent or financial decisions, an unsupported connective deserves more scrutiny than an ordinary style improvement. This scenario’s genre is operations log, and the source meaning brief is A potentially expected improvement did not appear; causal efficacy is not claimed. That is the benchmark for both the accuracy review and the final plain-language edit.

Case 27D. Evidence and author-query gate

An efficient reviewer does not write ‘unclear’ and stop. Ask a question that can be answered with evidence: Were services rerouted or frequencies changed independently? Include the source passage and the two candidate target readings. An author can then confirm the relation, provide missing context or explicitly leave the ambiguity open. The record should distinguish an author-confirmed meaning from a translator’s plausible interpretation and from a machine-generated suggestion.

After the answer, update the relation map, not only the visible sentence. If similar segments occur elsewhere, check them for consistency without mechanically propagating the same connective. Context may legitimately make the same pair of vocabulary items express a different relationship in a later document. For this exercise, the brief already supplies the acceptable decision: A potentially expected improvement did not appear; causal efficacy is not claimed. The next step is to verify that the chosen target preserves it and creates no new attribution.

Case 27E. Adversarial paraphrase and back-reading

Take the faithful candidate — “Despite the new track section opening, typical arrival times remained similar.” — and paraphrase it into a simple reader belief. Now take the unsafe candidate — “The new track made trains arrive slower.” — and do the same. Write down the extra proposition, presupposition or causal claim that appears in the second version. The target should not invent deterioration from the absence of improvement. This comparison is stronger than asking whether both sentences sound natural. A fluent sentence can be misleading, and a back-translation can repeat the same mistake if the reviewer already accepts its assumptions.

For an adversarial check, imagine a reader who accepts the two reported events but rejects the suggested link. Would that reader still be entitled to believe the source? Consider the possibility that opening capacity may affect resilience rather than speed. If yes, the stronger reading needs explicit textual or authorial support. The test does not require mathematical proof. It disciplines translation by making the asserted relationship visible and by preventing an intuition from hiding inside a connective.

Case 27F. Transfer drill and acceptance criterion

Transfer the same reasoning to another language, medium or genre without carrying over English word order. Keep the propositional facts and the stated relation fixed, but choose a natural construction suitable for the target audience. The final acceptable solution does not have to use the English connector in “Despite the new track section opening, typical arrival times remained similar.” literally. It does have to preserve a potentially expected improvement did not appear; causal efficacy is not claimed. It must also avoid the new belief introduced by “The new track made trains arrive slower.”.

The review is passed when a target reader can identify the two facts, draw only the source-authorized inference, and answer the practical question raised by this operations log text without adopting an unsupported explanation. The specific failure warning is: Decision makers might misjudge an investment. Keep the approval record with this sentence’s context, the reviewer’s decision and any author response. If a future revision changes the source relation, rerun this check rather than treating the existing translation as automatically safe.

Case 28: Journal review and formatting — Editorial workflow

Constructed source meaning: Reviewers recommended acceptance. The author still needed to fix figure labels.

Intended meaning supplied by the exercise brief: Scientific acceptance recommendation and production readiness are distinct milestones. Candidate that fails the brief: “The paper was fully published once reviewers approved it.” Defensible rendering under the brief: “Reviewers recommended acceptance, but figure labels still needed revision.”

Case 28A. Proposition isolation: what is actually stated

In this editorial workflow example, begin with the two source propositions without merging them into a single story: Reviewers recommended acceptance. The author still needed to fix figure labels. The first analytical task is to preserve each proposition and its evidential status. The passage is not a licence to invent a hidden third event, an extra agent or an unreported decision. A recommendation is not final publication or an unconditional approval. A translator who starts by choosing a smooth connective may overlook that the connective itself carries a claim. Start by marking A and B, their subjects and their temporal or logical scope; then describe the relationship separately in an interpretation note. The relevant relation in the approved training brief is: Scientific acceptance recommendation and production readiness are distinct milestones.

A useful minimality test is to ask whether each standalone sentence remains true if the proposed relation is false. Editorial and production steps may overlap. That alternative demonstrates why the words on the page and the link between them require separate auditing. Only the brief for this constructed case resolves the intended link. In an actual translation, identify what language, adjacent document context or author confirmation performs the equivalent role. Do not claim certainty merely because an explanation seems intuitive.

Case 28B. Competing reading and explicitness threshold

Compare the unsafe target, “The paper was fully published once reviewers approved it.”, with “Reviewers recommended acceptance, but figure labels still needed revision.”. They are not two equally faithful stylistic options under the stated brief. The former introduces a relationship or implication that needs independent justification; the latter retains the intended boundary. A recommendation is not final publication or an unconditional approval. Test the exact meaning of the linking expression, not merely the dictionary meanings of the nouns and verbs. Ask if the target asserts causation, expectation violation, permission, sequence, evidential proof, or exclusivity. If it does, locate the corresponding source commitment.

Where the real document is genuinely ambiguous, there are three defensible editorial routes. Preserve the ambiguity through natural target phrasing, seek a qualified source-side answer, or use an approved translator note that separates explanation from the translated claim. Do not silently invent a causal explanation to make the paragraph feel complete. The specific competing possibility here is: Editorial and production steps may overlap. That possibility may or may not be correct, but its continued logical availability shows why an overconfident connective can be dangerous.

Case 28C. Reader-effect and risk diagnostic

A relation error is not just a linguistic blemish; it changes what a reader is licensed to conclude. In this case: An author might skip required production tasks. Review the target as a person relying on the document rather than as an editor looking for elegant wording. Which action, inference, expectation or accusation becomes more likely after the bad version? What would the intended audience believe if the first source proposition were true but the inferred causal or concessive link were false?

A practical severity review should record the precise sentence, the changed inference, the intended reader, and the possible consequence. If the sentence affects safety, rights, access, technical diagnosis, consent or financial decisions, an unsupported connective deserves more scrutiny than an ordinary style improvement. This scenario’s genre is editorial workflow, and the source meaning brief is Scientific acceptance recommendation and production readiness are distinct milestones. That is the benchmark for both the accuracy review and the final plain-language edit.

Case 28D. Evidence and author-query gate

An efficient reviewer does not write ‘unclear’ and stop. Ask a question that can be answered with evidence: Was the editor’s final decision already issued? Include the source passage and the two candidate target readings. An author can then confirm the relation, provide missing context or explicitly leave the ambiguity open. The record should distinguish an author-confirmed meaning from a translator’s plausible interpretation and from a machine-generated suggestion.

After the answer, update the relation map, not only the visible sentence. If similar segments occur elsewhere, check them for consistency without mechanically propagating the same connective. Context may legitimately make the same pair of vocabulary items express a different relationship in a later document. For this exercise, the brief already supplies the acceptable decision: Scientific acceptance recommendation and production readiness are distinct milestones. The next step is to verify that the chosen target preserves it and creates no new attribution.

Case 28E. Adversarial paraphrase and back-reading

Take the faithful candidate — “Reviewers recommended acceptance, but figure labels still needed revision.” — and paraphrase it into a simple reader belief. Now take the unsafe candidate — “The paper was fully published once reviewers approved it.” — and do the same. Write down the extra proposition, presupposition or causal claim that appears in the second version. A recommendation is not final publication or an unconditional approval. This comparison is stronger than asking whether both sentences sound natural. A fluent sentence can be misleading, and a back-translation can repeat the same mistake if the reviewer already accepts its assumptions.

For an adversarial check, imagine a reader who accepts the two reported events but rejects the suggested link. Would that reader still be entitled to believe the source? Consider the possibility that editorial and production steps may overlap. If yes, the stronger reading needs explicit textual or authorial support. The test does not require mathematical proof. It disciplines translation by making the asserted relationship visible and by preventing an intuition from hiding inside a connective.

Case 28F. Transfer drill and acceptance criterion

Transfer the same reasoning to another language, medium or genre without carrying over English word order. Keep the propositional facts and the stated relation fixed, but choose a natural construction suitable for the target audience. The final acceptable solution does not have to use the English connector in “Reviewers recommended acceptance, but figure labels still needed revision.” literally. It does have to preserve scientific acceptance recommendation and production readiness are distinct milestones. It must also avoid the new belief introduced by “The paper was fully published once reviewers approved it.”.

The review is passed when a target reader can identify the two facts, draw only the source-authorized inference, and answer the practical question raised by this editorial workflow text without adopting an unsupported explanation. The specific failure warning is: An author might skip required production tasks. Keep the approval record with this sentence’s context, the reviewer’s decision and any author response. If a future revision changes the source relation, rerun this check rather than treating the existing translation as automatically safe.

Case 29: Cold storage and arrival — Logistics note

Constructed source meaning: The samples were refrigerated. They reached the laboratory the next afternoon.

Intended meaning supplied by the exercise brief: Handling and arrival sequence are recorded; no quality guarantee is asserted. Candidate that fails the brief: “The samples were viable because they were refrigerated.” Defensible rendering under the brief: “The samples were kept refrigerated and arrived the next afternoon.”

Case 29A. Proposition isolation: what is actually stated

In this logistics note example, begin with the two source propositions without merging them into a single story: The samples were refrigerated. They reached the laboratory the next afternoon. The first analytical task is to preserve each proposition and its evidential status. The passage is not a licence to invent a hidden third event, an extra agent or an unreported decision. Temperature handling alone cannot prove analytical validity without other data. A translator who starts by choosing a smooth connective may overlook that the connective itself carries a claim. Start by marking A and B, their subjects and their temporal or logical scope; then describe the relationship separately in an interpretation note. The relevant relation in the approved training brief is: Handling and arrival sequence are recorded; no quality guarantee is asserted.

A useful minimality test is to ask whether each standalone sentence remains true if the proposed relation is false. A test result may still be invalid for other reasons. That alternative demonstrates why the words on the page and the link between them require separate auditing. Only the brief for this constructed case resolves the intended link. In an actual translation, identify what language, adjacent document context or author confirmation performs the equivalent role. Do not claim certainty merely because an explanation seems intuitive.

Case 29B. Competing reading and explicitness threshold

Compare the unsafe target, “The samples were viable because they were refrigerated.”, with “The samples were kept refrigerated and arrived the next afternoon.”. They are not two equally faithful stylistic options under the stated brief. The former introduces a relationship or implication that needs independent justification; the latter retains the intended boundary. Temperature handling alone cannot prove analytical validity without other data. Test the exact meaning of the linking expression, not merely the dictionary meanings of the nouns and verbs. Ask if the target asserts causation, expectation violation, permission, sequence, evidential proof, or exclusivity. If it does, locate the corresponding source commitment.

Where the real document is genuinely ambiguous, there are three defensible editorial routes. Preserve the ambiguity through natural target phrasing, seek a qualified source-side answer, or use an approved translator note that separates explanation from the translated claim. Do not silently invent a causal explanation to make the paragraph feel complete. The specific competing possibility here is: A test result may still be invalid for other reasons. That possibility may or may not be correct, but its continued logical availability shows why an overconfident connective can be dangerous.

Case 29C. Reader-effect and risk diagnostic

A relation error is not just a linguistic blemish; it changes what a reader is licensed to conclude. In this case: A report reader may assume quality-control acceptance. Review the target as a person relying on the document rather than as an editor looking for elegant wording. Which action, inference, expectation or accusation becomes more likely after the bad version? What would the intended audience believe if the first source proposition were true but the inferred causal or concessive link were false?

A practical severity review should record the precise sentence, the changed inference, the intended reader, and the possible consequence. If the sentence affects safety, rights, access, technical diagnosis, consent or financial decisions, an unsupported connective deserves more scrutiny than an ordinary style improvement. This scenario’s genre is logistics note, and the source meaning brief is Handling and arrival sequence are recorded; no quality guarantee is asserted. That is the benchmark for both the accuracy review and the final plain-language edit.

Case 29D. Evidence and author-query gate

An efficient reviewer does not write ‘unclear’ and stop. Ask a question that can be answered with evidence: Were logs reviewed and temperature excursions checked? Include the source passage and the two candidate target readings. An author can then confirm the relation, provide missing context or explicitly leave the ambiguity open. The record should distinguish an author-confirmed meaning from a translator’s plausible interpretation and from a machine-generated suggestion.

After the answer, update the relation map, not only the visible sentence. If similar segments occur elsewhere, check them for consistency without mechanically propagating the same connective. Context may legitimately make the same pair of vocabulary items express a different relationship in a later document. For this exercise, the brief already supplies the acceptable decision: Handling and arrival sequence are recorded; no quality guarantee is asserted. The next step is to verify that the chosen target preserves it and creates no new attribution.

Case 29E. Adversarial paraphrase and back-reading

Take the faithful candidate — “The samples were kept refrigerated and arrived the next afternoon.” — and paraphrase it into a simple reader belief. Now take the unsafe candidate — “The samples were viable because they were refrigerated.” — and do the same. Write down the extra proposition, presupposition or causal claim that appears in the second version. Temperature handling alone cannot prove analytical validity without other data. This comparison is stronger than asking whether both sentences sound natural. A fluent sentence can be misleading, and a back-translation can repeat the same mistake if the reviewer already accepts its assumptions.

For an adversarial check, imagine a reader who accepts the two reported events but rejects the suggested link. Would that reader still be entitled to believe the source? Consider the possibility that a test result may still be invalid for other reasons. If yes, the stronger reading needs explicit textual or authorial support. The test does not require mathematical proof. It disciplines translation by making the asserted relationship visible and by preventing an intuition from hiding inside a connective.

Case 29F. Transfer drill and acceptance criterion

Transfer the same reasoning to another language, medium or genre without carrying over English word order. Keep the propositional facts and the stated relation fixed, but choose a natural construction suitable for the target audience. The final acceptable solution does not have to use the English connector in “The samples were kept refrigerated and arrived the next afternoon.” literally. It does have to preserve handling and arrival sequence are recorded; no quality guarantee is asserted. It must also avoid the new belief introduced by “The samples were viable because they were refrigerated.”.

The review is passed when a target reader can identify the two facts, draw only the source-authorized inference, and answer the practical question raised by this logistics note text without adopting an unsupported explanation. The specific failure warning is: A report reader may assume quality-control acceptance. Keep the approval record with this sentence’s context, the reviewer’s decision and any author response. If a future revision changes the source relation, rerun this check rather than treating the existing translation as automatically safe.

Case 30: A hearing and a postponement — Civic agenda

Constructed source meaning: The hearing opened. The final vote was postponed.

Intended meaning supplied by the exercise brief: A meeting stage proceeded while a later decision stage did not. Candidate that fails the brief: “The entire hearing was cancelled before it began.” Defensible rendering under the brief: “The hearing took place, but the final vote was postponed.”

Case 30A. Proposition isolation: what is actually stated

In this civic agenda example, begin with the two source propositions without merging them into a single story: The hearing opened. The final vote was postponed. The first analytical task is to preserve each proposition and its evidential status. The passage is not a licence to invent a hidden third event, an extra agent or an unreported decision. A stage-specific change must not become a blanket cancellation. A translator who starts by choosing a smooth connective may overlook that the connective itself carries a claim. Start by marking A and B, their subjects and their temporal or logical scope; then describe the relationship separately in an interpretation note. The relevant relation in the approved training brief is: A meeting stage proceeded while a later decision stage did not.

A useful minimality test is to ask whether each standalone sentence remains true if the proposed relation is false. Additional evidence might be the reason for postponement. That alternative demonstrates why the words on the page and the link between them require separate auditing. Only the brief for this constructed case resolves the intended link. In an actual translation, identify what language, adjacent document context or author confirmation performs the equivalent role. Do not claim certainty merely because an explanation seems intuitive.

Case 30B. Competing reading and explicitness threshold

Compare the unsafe target, “The entire hearing was cancelled before it began.”, with “The hearing took place, but the final vote was postponed.”. They are not two equally faithful stylistic options under the stated brief. The former introduces a relationship or implication that needs independent justification; the latter retains the intended boundary. A stage-specific change must not become a blanket cancellation. Test the exact meaning of the linking expression, not merely the dictionary meanings of the nouns and verbs. Ask if the target asserts causation, expectation violation, permission, sequence, evidential proof, or exclusivity. If it does, locate the corresponding source commitment.

Where the real document is genuinely ambiguous, there are three defensible editorial routes. Preserve the ambiguity through natural target phrasing, seek a qualified source-side answer, or use an approved translator note that separates explanation from the translated claim. Do not silently invent a causal explanation to make the paragraph feel complete. The specific competing possibility here is: Additional evidence might be the reason for postponement. That possibility may or may not be correct, but its continued logical availability shows why an overconfident connective can be dangerous.

Case 30C. Reader-effect and risk diagnostic

A relation error is not just a linguistic blemish; it changes what a reader is licensed to conclude. In this case: Participants may believe testimony was never heard. Review the target as a person relying on the document rather than as an editor looking for elegant wording. Which action, inference, expectation or accusation becomes more likely after the bad version? What would the intended audience believe if the first source proposition were true but the inferred causal or concessive link were false?

A practical severity review should record the precise sentence, the changed inference, the intended reader, and the possible consequence. If the sentence affects safety, rights, access, technical diagnosis, consent or financial decisions, an unsupported connective deserves more scrutiny than an ordinary style improvement. This scenario’s genre is civic agenda, and the source meaning brief is A meeting stage proceeded while a later decision stage did not. That is the benchmark for both the accuracy review and the final plain-language edit.

Case 30D. Evidence and author-query gate

An efficient reviewer does not write ‘unclear’ and stop. Ask a question that can be answered with evidence: Was the vote postponed to a fixed date? Include the source passage and the two candidate target readings. An author can then confirm the relation, provide missing context or explicitly leave the ambiguity open. The record should distinguish an author-confirmed meaning from a translator’s plausible interpretation and from a machine-generated suggestion.

After the answer, update the relation map, not only the visible sentence. If similar segments occur elsewhere, check them for consistency without mechanically propagating the same connective. Context may legitimately make the same pair of vocabulary items express a different relationship in a later document. For this exercise, the brief already supplies the acceptable decision: A meeting stage proceeded while a later decision stage did not. The next step is to verify that the chosen target preserves it and creates no new attribution.

Case 30E. Adversarial paraphrase and back-reading

Take the faithful candidate — “The hearing took place, but the final vote was postponed.” — and paraphrase it into a simple reader belief. Now take the unsafe candidate — “The entire hearing was cancelled before it began.” — and do the same. Write down the extra proposition, presupposition or causal claim that appears in the second version. A stage-specific change must not become a blanket cancellation. This comparison is stronger than asking whether both sentences sound natural. A fluent sentence can be misleading, and a back-translation can repeat the same mistake if the reviewer already accepts its assumptions.

For an adversarial check, imagine a reader who accepts the two reported events but rejects the suggested link. Would that reader still be entitled to believe the source? Consider the possibility that additional evidence might be the reason for postponement. If yes, the stronger reading needs explicit textual or authorial support. The test does not require mathematical proof. It disciplines translation by making the asserted relationship visible and by preventing an intuition from hiding inside a connective.

Case 30F. Transfer drill and acceptance criterion

Transfer the same reasoning to another language, medium or genre without carrying over English word order. Keep the propositional facts and the stated relation fixed, but choose a natural construction suitable for the target audience. The final acceptable solution does not have to use the English connector in “The hearing took place, but the final vote was postponed.” literally. It does have to preserve a meeting stage proceeded while a later decision stage did not. It must also avoid the new belief introduced by “The entire hearing was cancelled before it began.”.

The review is passed when a target reader can identify the two facts, draw only the source-authorized inference, and answer the practical question raised by this civic agenda text without adopting an unsupported explanation. The specific failure warning is: Participants may believe testimony was never heard. Keep the approval record with this sentence’s context, the reviewer’s decision and any author response. If a future revision changes the source relation, rerun this check rather than treating the existing translation as automatically safe.

Case 31: Failed test and cable check — Engineering ticket

Constructed source meaning: The trial did not pass. An engineer checked the cable.

Intended meaning supplied by the exercise brief: The cable check followed failure as diagnostic work; no root cause is confirmed. Candidate that fails the brief: “The test failed because the cable was faulty.” Defensible rendering under the brief: “After the test failed, an engineer checked the cable.”

Case 31A. Proposition isolation: what is actually stated

In this engineering ticket example, begin with the two source propositions without merging them into a single story: The trial did not pass. An engineer checked the cable. The first analytical task is to preserve each proposition and its evidential status. The passage is not a licence to invent a hidden third event, an extra agent or an unreported decision. Investigation of a component is not proof that component caused the failure. A translator who starts by choosing a smooth connective may overlook that the connective itself carries a claim. Start by marking A and B, their subjects and their temporal or logical scope; then describe the relationship separately in an interpretation note. The relevant relation in the approved training brief is: The cable check followed failure as diagnostic work; no root cause is confirmed.

A useful minimality test is to ask whether each standalone sentence remains true if the proposed relation is false. The failure could originate in software, power or procedure. That alternative demonstrates why the words on the page and the link between them require separate auditing. Only the brief for this constructed case resolves the intended link. In an actual translation, identify what language, adjacent document context or author confirmation performs the equivalent role. Do not claim certainty merely because an explanation seems intuitive.

Case 31B. Competing reading and explicitness threshold

Compare the unsafe target, “The test failed because the cable was faulty.”, with “After the test failed, an engineer checked the cable.”. They are not two equally faithful stylistic options under the stated brief. The former introduces a relationship or implication that needs independent justification; the latter retains the intended boundary. Investigation of a component is not proof that component caused the failure. Test the exact meaning of the linking expression, not merely the dictionary meanings of the nouns and verbs. Ask if the target asserts causation, expectation violation, permission, sequence, evidential proof, or exclusivity. If it does, locate the corresponding source commitment.

Where the real document is genuinely ambiguous, there are three defensible editorial routes. Preserve the ambiguity through natural target phrasing, seek a qualified source-side answer, or use an approved translator note that separates explanation from the translated claim. Do not silently invent a causal explanation to make the paragraph feel complete. The specific competing possibility here is: The failure could originate in software, power or procedure. That possibility may or may not be correct, but its continued logical availability shows why an overconfident connective can be dangerous.

Case 31C. Reader-effect and risk diagnostic

A relation error is not just a linguistic blemish; it changes what a reader is licensed to conclude. In this case: A service report might unfairly assign fault. Review the target as a person relying on the document rather than as an editor looking for elegant wording. Which action, inference, expectation or accusation becomes more likely after the bad version? What would the intended audience believe if the first source proposition were true but the inferred causal or concessive link were false?

A practical severity review should record the precise sentence, the changed inference, the intended reader, and the possible consequence. If the sentence affects safety, rights, access, technical diagnosis, consent or financial decisions, an unsupported connective deserves more scrutiny than an ordinary style improvement. This scenario’s genre is engineering ticket, and the source meaning brief is The cable check followed failure as diagnostic work; no root cause is confirmed. That is the benchmark for both the accuracy review and the final plain-language edit.

Case 31D. Evidence and author-query gate

An efficient reviewer does not write ‘unclear’ and stop. Ask a question that can be answered with evidence: Did subsequent testing demonstrate a cable defect? Include the source passage and the two candidate target readings. An author can then confirm the relation, provide missing context or explicitly leave the ambiguity open. The record should distinguish an author-confirmed meaning from a translator’s plausible interpretation and from a machine-generated suggestion.

After the answer, update the relation map, not only the visible sentence. If similar segments occur elsewhere, check them for consistency without mechanically propagating the same connective. Context may legitimately make the same pair of vocabulary items express a different relationship in a later document. For this exercise, the brief already supplies the acceptable decision: The cable check followed failure as diagnostic work; no root cause is confirmed. The next step is to verify that the chosen target preserves it and creates no new attribution.

Case 31E. Adversarial paraphrase and back-reading

Take the faithful candidate — “After the test failed, an engineer checked the cable.” — and paraphrase it into a simple reader belief. Now take the unsafe candidate — “The test failed because the cable was faulty.” — and do the same. Write down the extra proposition, presupposition or causal claim that appears in the second version. Investigation of a component is not proof that component caused the failure. This comparison is stronger than asking whether both sentences sound natural. A fluent sentence can be misleading, and a back-translation can repeat the same mistake if the reviewer already accepts its assumptions.

For an adversarial check, imagine a reader who accepts the two reported events but rejects the suggested link. Would that reader still be entitled to believe the source? Consider the possibility that the failure could originate in software, power or procedure. If yes, the stronger reading needs explicit textual or authorial support. The test does not require mathematical proof. It disciplines translation by making the asserted relationship visible and by preventing an intuition from hiding inside a connective.

Case 31F. Transfer drill and acceptance criterion

Transfer the same reasoning to another language, medium or genre without carrying over English word order. Keep the propositional facts and the stated relation fixed, but choose a natural construction suitable for the target audience. The final acceptable solution does not have to use the English connector in “After the test failed, an engineer checked the cable.” literally. It does have to preserve the cable check followed failure as diagnostic work; no root cause is confirmed. It must also avoid the new belief introduced by “The test failed because the cable was faulty.”.

The review is passed when a target reader can identify the two facts, draw only the source-authorized inference, and answer the practical question raised by this engineering ticket text without adopting an unsupported explanation. The specific failure warning is: A service report might unfairly assign fault. Keep the approval record with this sentence’s context, the reviewer’s decision and any author response. If a future revision changes the source relation, rerun this check rather than treating the existing translation as automatically safe.

Case 32: Limited publication rights — Archive request

Constructed source meaning: The image can be viewed online. Reproduction requires permission.

Intended meaning supplied by the exercise brief: Access and reuse are different rights; the second clause is a restriction. Candidate that fails the brief: “Because the image is online, it may be reproduced freely.” Defensible rendering under the brief: “The image is available to view online, but reproduction requires permission.”

Case 32A. Proposition isolation: what is actually stated

In this archive request example, begin with the two source propositions without merging them into a single story: The image can be viewed online. Reproduction requires permission. The first analytical task is to preserve each proposition and its evidential status. The passage is not a licence to invent a hidden third event, an extra agent or an unreported decision. Ease of viewing does not establish licensing rights. A translator who starts by choosing a smooth connective may overlook that the connective itself carries a claim. Start by marking A and B, their subjects and their temporal or logical scope; then describe the relationship separately in an interpretation note. The relevant relation in the approved training brief is: Access and reuse are different rights; the second clause is a restriction.

A useful minimality test is to ask whether each standalone sentence remains true if the proposed relation is false. The image may have multiple overlapping rights holders. That alternative demonstrates why the words on the page and the link between them require separate auditing. Only the brief for this constructed case resolves the intended link. In an actual translation, identify what language, adjacent document context or author confirmation performs the equivalent role. Do not claim certainty merely because an explanation seems intuitive.

Case 32B. Competing reading and explicitness threshold

Compare the unsafe target, “Because the image is online, it may be reproduced freely.”, with “The image is available to view online, but reproduction requires permission.”. They are not two equally faithful stylistic options under the stated brief. The former introduces a relationship or implication that needs independent justification; the latter retains the intended boundary. Ease of viewing does not establish licensing rights. Test the exact meaning of the linking expression, not merely the dictionary meanings of the nouns and verbs. Ask if the target asserts causation, expectation violation, permission, sequence, evidential proof, or exclusivity. If it does, locate the corresponding source commitment.

Where the real document is genuinely ambiguous, there are three defensible editorial routes. Preserve the ambiguity through natural target phrasing, seek a qualified source-side answer, or use an approved translator note that separates explanation from the translated claim. Do not silently invent a causal explanation to make the paragraph feel complete. The specific competing possibility here is: The image may have multiple overlapping rights holders. That possibility may or may not be correct, but its continued logical availability shows why an overconfident connective can be dangerous.

Case 32C. Reader-effect and risk diagnostic

A relation error is not just a linguistic blemish; it changes what a reader is licensed to conclude. In this case: Readers could copy material without the required authorization. Review the target as a person relying on the document rather than as an editor looking for elegant wording. Which action, inference, expectation or accusation becomes more likely after the bad version? What would the intended audience believe if the first source proposition were true but the inferred causal or concessive link were false?

A practical severity review should record the precise sentence, the changed inference, the intended reader, and the possible consequence. If the sentence affects safety, rights, access, technical diagnosis, consent or financial decisions, an unsupported connective deserves more scrutiny than an ordinary style improvement. This scenario’s genre is archive request, and the source meaning brief is Access and reuse are different rights; the second clause is a restriction. That is the benchmark for both the accuracy review and the final plain-language edit.

Case 32D. Evidence and author-query gate

An efficient reviewer does not write ‘unclear’ and stop. Ask a question that can be answered with evidence: Is quotation covered by a separate exception in the governing policy? Include the source passage and the two candidate target readings. An author can then confirm the relation, provide missing context or explicitly leave the ambiguity open. The record should distinguish an author-confirmed meaning from a translator’s plausible interpretation and from a machine-generated suggestion.

After the answer, update the relation map, not only the visible sentence. If similar segments occur elsewhere, check them for consistency without mechanically propagating the same connective. Context may legitimately make the same pair of vocabulary items express a different relationship in a later document. For this exercise, the brief already supplies the acceptable decision: Access and reuse are different rights; the second clause is a restriction. The next step is to verify that the chosen target preserves it and creates no new attribution.

Case 32E. Adversarial paraphrase and back-reading

Take the faithful candidate — “The image is available to view online, but reproduction requires permission.” — and paraphrase it into a simple reader belief. Now take the unsafe candidate — “Because the image is online, it may be reproduced freely.” — and do the same. Write down the extra proposition, presupposition or causal claim that appears in the second version. Ease of viewing does not establish licensing rights. This comparison is stronger than asking whether both sentences sound natural. A fluent sentence can be misleading, and a back-translation can repeat the same mistake if the reviewer already accepts its assumptions.

For an adversarial check, imagine a reader who accepts the two reported events but rejects the suggested link. Would that reader still be entitled to believe the source? Consider the possibility that the image may have multiple overlapping rights holders. If yes, the stronger reading needs explicit textual or authorial support. The test does not require mathematical proof. It disciplines translation by making the asserted relationship visible and by preventing an intuition from hiding inside a connective.

Case 32F. Transfer drill and acceptance criterion

Transfer the same reasoning to another language, medium or genre without carrying over English word order. Keep the propositional facts and the stated relation fixed, but choose a natural construction suitable for the target audience. The final acceptable solution does not have to use the English connector in “The image is available to view online, but reproduction requires permission.” literally. It does have to preserve access and reuse are different rights; the second clause is a restriction. It must also avoid the new belief introduced by “Because the image is online, it may be reproduced freely.”.

The review is passed when a target reader can identify the two facts, draw only the source-authorized inference, and answer the practical question raised by this archive request text without adopting an unsupported explanation. The specific failure warning is: Readers could copy material without the required authorization. Keep the approval record with this sentence’s context, the reviewer’s decision and any author response. If a future revision changes the source relation, rerun this check rather than treating the existing translation as automatically safe.

Case 33: Two sensor channels — Instrument manual

Constructed source meaning: Channel A measures temperature. Channel B measures humidity.

Intended meaning supplied by the exercise brief: Parallel functional descriptions; no superiority or sequential dependence is intended. Candidate that fails the brief: “Channel A measures temperature so that channel B can measure humidity.” Defensible rendering under the brief: “Channel A measures temperature; channel B measures humidity.”

Case 33A. Proposition isolation: what is actually stated

In this instrument manual example, begin with the two source propositions without merging them into a single story: Channel A measures temperature. Channel B measures humidity. The first analytical task is to preserve each proposition and its evidential status. The passage is not a licence to invent a hidden third event, an extra agent or an unreported decision. Putting purpose between independent channels invents a design dependency. A translator who starts by choosing a smooth connective may overlook that the connective itself carries a claim. Start by marking A and B, their subjects and their temporal or logical scope; then describe the relationship separately in an interpretation note. The relevant relation in the approved training brief is: Parallel functional descriptions; no superiority or sequential dependence is intended.

A useful minimality test is to ask whether each standalone sentence remains true if the proposed relation is false. A multi-sensor module may still share components. That alternative demonstrates why the words on the page and the link between them require separate auditing. Only the brief for this constructed case resolves the intended link. In an actual translation, identify what language, adjacent document context or author confirmation performs the equivalent role. Do not claim certainty merely because an explanation seems intuitive.

Case 33B. Competing reading and explicitness threshold

Compare the unsafe target, “Channel A measures temperature so that channel B can measure humidity.”, with “Channel A measures temperature; channel B measures humidity.”. They are not two equally faithful stylistic options under the stated brief. The former introduces a relationship or implication that needs independent justification; the latter retains the intended boundary. Putting purpose between independent channels invents a design dependency. Test the exact meaning of the linking expression, not merely the dictionary meanings of the nouns and verbs. Ask if the target asserts causation, expectation violation, permission, sequence, evidential proof, or exclusivity. If it does, locate the corresponding source commitment.

Where the real document is genuinely ambiguous, there are three defensible editorial routes. Preserve the ambiguity through natural target phrasing, seek a qualified source-side answer, or use an approved translator note that separates explanation from the translated claim. Do not silently invent a causal explanation to make the paragraph feel complete. The specific competing possibility here is: A multi-sensor module may still share components. That possibility may or may not be correct, but its continued logical availability shows why an overconfident connective can be dangerous.

Case 33C. Reader-effect and risk diagnostic

A relation error is not just a linguistic blemish; it changes what a reader is licensed to conclude. In this case: A technician could diagnose the wrong subsystem. Review the target as a person relying on the document rather than as an editor looking for elegant wording. Which action, inference, expectation or accusation becomes more likely after the bad version? What would the intended audience believe if the first source proposition were true but the inferred causal or concessive link were false?

A practical severity review should record the precise sentence, the changed inference, the intended reader, and the possible consequence. If the sentence affects safety, rights, access, technical diagnosis, consent or financial decisions, an unsupported connective deserves more scrutiny than an ordinary style improvement. This scenario’s genre is instrument manual, and the source meaning brief is Parallel functional descriptions; no superiority or sequential dependence is intended. That is the benchmark for both the accuracy review and the final plain-language edit.

Case 33D. Evidence and author-query gate

An efficient reviewer does not write ‘unclear’ and stop. Ask a question that can be answered with evidence: Are the channels electrically independent in this model? Include the source passage and the two candidate target readings. An author can then confirm the relation, provide missing context or explicitly leave the ambiguity open. The record should distinguish an author-confirmed meaning from a translator’s plausible interpretation and from a machine-generated suggestion.

After the answer, update the relation map, not only the visible sentence. If similar segments occur elsewhere, check them for consistency without mechanically propagating the same connective. Context may legitimately make the same pair of vocabulary items express a different relationship in a later document. For this exercise, the brief already supplies the acceptable decision: Parallel functional descriptions; no superiority or sequential dependence is intended. The next step is to verify that the chosen target preserves it and creates no new attribution.

Case 33E. Adversarial paraphrase and back-reading

Take the faithful candidate — “Channel A measures temperature; channel B measures humidity.” — and paraphrase it into a simple reader belief. Now take the unsafe candidate — “Channel A measures temperature so that channel B can measure humidity.” — and do the same. Write down the extra proposition, presupposition or causal claim that appears in the second version. Putting purpose between independent channels invents a design dependency. This comparison is stronger than asking whether both sentences sound natural. A fluent sentence can be misleading, and a back-translation can repeat the same mistake if the reviewer already accepts its assumptions.

For an adversarial check, imagine a reader who accepts the two reported events but rejects the suggested link. Would that reader still be entitled to believe the source? Consider the possibility that a multi-sensor module may still share components. If yes, the stronger reading needs explicit textual or authorial support. The test does not require mathematical proof. It disciplines translation by making the asserted relationship visible and by preventing an intuition from hiding inside a connective.

Case 33F. Transfer drill and acceptance criterion

Transfer the same reasoning to another language, medium or genre without carrying over English word order. Keep the propositional facts and the stated relation fixed, but choose a natural construction suitable for the target audience. The final acceptable solution does not have to use the English connector in “Channel A measures temperature; channel B measures humidity.” literally. It does have to preserve parallel functional descriptions; no superiority or sequential dependence is intended. It must also avoid the new belief introduced by “Channel A measures temperature so that channel B can measure humidity.”.

The review is passed when a target reader can identify the two facts, draw only the source-authorized inference, and answer the practical question raised by this instrument manual text without adopting an unsupported explanation. The specific failure warning is: A technician could diagnose the wrong subsystem. Keep the approval record with this sentence’s context, the reviewer’s decision and any author response. If a future revision changes the source relation, rerun this check rather than treating the existing translation as automatically safe.

Case 34: Public debate and evidence — News analysis

Constructed source meaning: One group welcomed the proposal. Another asked for more evidence.

Intended meaning supplied by the exercise brief: The second reaction is an independent position, not proof the first group ignored evidence. Candidate that fails the brief: “The supporters welcomed the proposal despite knowing it had no evidence.” Defensible rendering under the brief: “One group welcomed the proposal; another requested additional evidence.”

Case 34A. Proposition isolation: what is actually stated

In this news analysis example, begin with the two source propositions without merging them into a single story: One group welcomed the proposal. Another asked for more evidence. The first analytical task is to preserve each proposition and its evidential status. The passage is not a licence to invent a hidden third event, an extra agent or an unreported decision. The faulty version attributes knowledge and motive to supporters without support. A translator who starts by choosing a smooth connective may overlook that the connective itself carries a claim. Start by marking A and B, their subjects and their temporal or logical scope; then describe the relationship separately in an interpretation note. The relevant relation in the approved training brief is: The second reaction is an independent position, not proof the first group ignored evidence.

A useful minimality test is to ask whether each standalone sentence remains true if the proposed relation is false. Different groups could be prioritizing different concerns. That alternative demonstrates why the words on the page and the link between them require separate auditing. Only the brief for this constructed case resolves the intended link. In an actual translation, identify what language, adjacent document context or author confirmation performs the equivalent role. Do not claim certainty merely because an explanation seems intuitive.

Case 34B. Competing reading and explicitness threshold

Compare the unsafe target, “The supporters welcomed the proposal despite knowing it had no evidence.”, with “One group welcomed the proposal; another requested additional evidence.”. They are not two equally faithful stylistic options under the stated brief. The former introduces a relationship or implication that needs independent justification; the latter retains the intended boundary. The faulty version attributes knowledge and motive to supporters without support. Test the exact meaning of the linking expression, not merely the dictionary meanings of the nouns and verbs. Ask if the target asserts causation, expectation violation, permission, sequence, evidential proof, or exclusivity. If it does, locate the corresponding source commitment.

Where the real document is genuinely ambiguous, there are three defensible editorial routes. Preserve the ambiguity through natural target phrasing, seek a qualified source-side answer, or use an approved translator note that separates explanation from the translated claim. Do not silently invent a causal explanation to make the paragraph feel complete. The specific competing possibility here is: Different groups could be prioritizing different concerns. That possibility may or may not be correct, but its continued logical availability shows why an overconfident connective can be dangerous.

Case 34C. Reader-effect and risk diagnostic

A relation error is not just a linguistic blemish; it changes what a reader is licensed to conclude. In this case: Readers may assign bad faith to named groups. Review the target as a person relying on the document rather than as an editor looking for elegant wording. Which action, inference, expectation or accusation becomes more likely after the bad version? What would the intended audience believe if the first source proposition were true but the inferred causal or concessive link were false?

A practical severity review should record the precise sentence, the changed inference, the intended reader, and the possible consequence. If the sentence affects safety, rights, access, technical diagnosis, consent or financial decisions, an unsupported connective deserves more scrutiny than an ordinary style improvement. This scenario’s genre is news analysis, and the source meaning brief is The second reaction is an independent position, not proof the first group ignored evidence. That is the benchmark for both the accuracy review and the final plain-language edit.

Case 34D. Evidence and author-query gate

An efficient reviewer does not write ‘unclear’ and stop. Ask a question that can be answered with evidence: Did any participant explicitly reject evidence or only request more? Include the source passage and the two candidate target readings. An author can then confirm the relation, provide missing context or explicitly leave the ambiguity open. The record should distinguish an author-confirmed meaning from a translator’s plausible interpretation and from a machine-generated suggestion.

After the answer, update the relation map, not only the visible sentence. If similar segments occur elsewhere, check them for consistency without mechanically propagating the same connective. Context may legitimately make the same pair of vocabulary items express a different relationship in a later document. For this exercise, the brief already supplies the acceptable decision: The second reaction is an independent position, not proof the first group ignored evidence. The next step is to verify that the chosen target preserves it and creates no new attribution.

Case 34E. Adversarial paraphrase and back-reading

Take the faithful candidate — “One group welcomed the proposal; another requested additional evidence.” — and paraphrase it into a simple reader belief. Now take the unsafe candidate — “The supporters welcomed the proposal despite knowing it had no evidence.” — and do the same. Write down the extra proposition, presupposition or causal claim that appears in the second version. The faulty version attributes knowledge and motive to supporters without support. This comparison is stronger than asking whether both sentences sound natural. A fluent sentence can be misleading, and a back-translation can repeat the same mistake if the reviewer already accepts its assumptions.

For an adversarial check, imagine a reader who accepts the two reported events but rejects the suggested link. Would that reader still be entitled to believe the source? Consider the possibility that different groups could be prioritizing different concerns. If yes, the stronger reading needs explicit textual or authorial support. The test does not require mathematical proof. It disciplines translation by making the asserted relationship visible and by preventing an intuition from hiding inside a connective.

Case 34F. Transfer drill and acceptance criterion

Transfer the same reasoning to another language, medium or genre without carrying over English word order. Keep the propositional facts and the stated relation fixed, but choose a natural construction suitable for the target audience. The final acceptable solution does not have to use the English connector in “One group welcomed the proposal; another requested additional evidence.” literally. It does have to preserve the second reaction is an independent position, not proof the first group ignored evidence. It must also avoid the new belief introduced by “The supporters welcomed the proposal despite knowing it had no evidence.”.

The review is passed when a target reader can identify the two facts, draw only the source-authorized inference, and answer the practical question raised by this news analysis text without adopting an unsupported explanation. The specific failure warning is: Readers may assign bad faith to named groups. Keep the approval record with this sentence’s context, the reviewer’s decision and any author response. If a future revision changes the source relation, rerun this check rather than treating the existing translation as automatically safe.

Case 35: Failed alert and manual response — Safety incident note

Constructed source meaning: The automatic alert did not trigger. A worker called the control room.

Intended meaning supplied by the exercise brief: The manual action followed a missing automatic signal; attribution of intention is not fully stated. Candidate that fails the brief: “The worker disabled the alert and then contacted control.” Defensible rendering under the brief: “The automatic alert did not trigger; a worker contacted the control room.”

Case 35A. Proposition isolation: what is actually stated

In this safety incident note example, begin with the two source propositions without merging them into a single story: The automatic alert did not trigger. A worker called the control room. The first analytical task is to preserve each proposition and its evidential status. The passage is not a licence to invent a hidden third event, an extra agent or an unreported decision. The mistaken reading assigns an intentional disabling act rather than failure to trigger. A translator who starts by choosing a smooth connective may overlook that the connective itself carries a claim. Start by marking A and B, their subjects and their temporal or logical scope; then describe the relationship separately in an interpretation note. The relevant relation in the approved training brief is: The manual action followed a missing automatic signal; attribution of intention is not fully stated.

A useful minimality test is to ask whether each standalone sentence remains true if the proposed relation is false. The manual call may have been prompted by independent observation. That alternative demonstrates why the words on the page and the link between them require separate auditing. Only the brief for this constructed case resolves the intended link. In an actual translation, identify what language, adjacent document context or author confirmation performs the equivalent role. Do not claim certainty merely because an explanation seems intuitive.

Case 35B. Competing reading and explicitness threshold

Compare the unsafe target, “The worker disabled the alert and then contacted control.”, with “The automatic alert did not trigger; a worker contacted the control room.”. They are not two equally faithful stylistic options under the stated brief. The former introduces a relationship or implication that needs independent justification; the latter retains the intended boundary. The mistaken reading assigns an intentional disabling act rather than failure to trigger. Test the exact meaning of the linking expression, not merely the dictionary meanings of the nouns and verbs. Ask if the target asserts causation, expectation violation, permission, sequence, evidential proof, or exclusivity. If it does, locate the corresponding source commitment.

Where the real document is genuinely ambiguous, there are three defensible editorial routes. Preserve the ambiguity through natural target phrasing, seek a qualified source-side answer, or use an approved translator note that separates explanation from the translated claim. Do not silently invent a causal explanation to make the paragraph feel complete. The specific competing possibility here is: The manual call may have been prompted by independent observation. That possibility may or may not be correct, but its continued logical availability shows why an overconfident connective can be dangerous.

Case 35C. Reader-effect and risk diagnostic

A relation error is not just a linguistic blemish; it changes what a reader is licensed to conclude. In this case: Investigators could misidentify the source of the incident. Review the target as a person relying on the document rather than as an editor looking for elegant wording. Which action, inference, expectation or accusation becomes more likely after the bad version? What would the intended audience believe if the first source proposition were true but the inferred causal or concessive link were false?

A practical severity review should record the precise sentence, the changed inference, the intended reader, and the possible consequence. If the sentence affects safety, rights, access, technical diagnosis, consent or financial decisions, an unsupported connective deserves more scrutiny than an ordinary style improvement. This scenario’s genre is safety incident note, and the source meaning brief is The manual action followed a missing automatic signal; attribution of intention is not fully stated. That is the benchmark for both the accuracy review and the final plain-language edit.

Case 35D. Evidence and author-query gate

An efficient reviewer does not write ‘unclear’ and stop. Ask a question that can be answered with evidence: Was the worker aware of the failed alert when calling? Include the source passage and the two candidate target readings. An author can then confirm the relation, provide missing context or explicitly leave the ambiguity open. The record should distinguish an author-confirmed meaning from a translator’s plausible interpretation and from a machine-generated suggestion.

After the answer, update the relation map, not only the visible sentence. If similar segments occur elsewhere, check them for consistency without mechanically propagating the same connective. Context may legitimately make the same pair of vocabulary items express a different relationship in a later document. For this exercise, the brief already supplies the acceptable decision: The manual action followed a missing automatic signal; attribution of intention is not fully stated. The next step is to verify that the chosen target preserves it and creates no new attribution.

Case 35E. Adversarial paraphrase and back-reading

Take the faithful candidate — “The automatic alert did not trigger; a worker contacted the control room.” — and paraphrase it into a simple reader belief. Now take the unsafe candidate — “The worker disabled the alert and then contacted control.” — and do the same. Write down the extra proposition, presupposition or causal claim that appears in the second version. The mistaken reading assigns an intentional disabling act rather than failure to trigger. This comparison is stronger than asking whether both sentences sound natural. A fluent sentence can be misleading, and a back-translation can repeat the same mistake if the reviewer already accepts its assumptions.

For an adversarial check, imagine a reader who accepts the two reported events but rejects the suggested link. Would that reader still be entitled to believe the source? Consider the possibility that the manual call may have been prompted by independent observation. If yes, the stronger reading needs explicit textual or authorial support. The test does not require mathematical proof. It disciplines translation by making the asserted relationship visible and by preventing an intuition from hiding inside a connective.

Case 35F. Transfer drill and acceptance criterion

Transfer the same reasoning to another language, medium or genre without carrying over English word order. Keep the propositional facts and the stated relation fixed, but choose a natural construction suitable for the target audience. The final acceptable solution does not have to use the English connector in “The automatic alert did not trigger; a worker contacted the control room.” literally. It does have to preserve the manual action followed a missing automatic signal; attribution of intention is not fully stated. It must also avoid the new belief introduced by “The worker disabled the alert and then contacted control.”.

The review is passed when a target reader can identify the two facts, draw only the source-authorized inference, and answer the practical question raised by this safety incident note text without adopting an unsupported explanation. The specific failure warning is: Investigators could misidentify the source of the incident. Keep the approval record with this sentence’s context, the reviewer’s decision and any author response. If a future revision changes the source relation, rerun this check rather than treating the existing translation as automatically safe.

Case 36: Survey completion and confidence — Public research

Constructed source meaning: Many respondents completed the survey. The result was still described as uncertain.

Intended meaning supplied by the exercise brief: Response volume does not alone establish representativeness or confidence. Candidate that fails the brief: “The many responses proved the result certain.” Defensible rendering under the brief: “Although many people responded, the result remained uncertain.”

Case 36A. Proposition isolation: what is actually stated

In this public research example, begin with the two source propositions without merging them into a single story: Many respondents completed the survey. The result was still described as uncertain. The first analytical task is to preserve each proposition and its evidential status. The passage is not a licence to invent a hidden third event, an extra agent or an unreported decision. A high count can coexist with selection bias or measurement limits. A translator who starts by choosing a smooth connective may overlook that the connective itself carries a claim. Start by marking A and B, their subjects and their temporal or logical scope; then describe the relationship separately in an interpretation note. The relevant relation in the approved training brief is: Response volume does not alone establish representativeness or confidence.

A useful minimality test is to ask whether each standalone sentence remains true if the proposed relation is false. Coverage errors could remain even with many responses. That alternative demonstrates why the words on the page and the link between them require separate auditing. Only the brief for this constructed case resolves the intended link. In an actual translation, identify what language, adjacent document context or author confirmation performs the equivalent role. Do not claim certainty merely because an explanation seems intuitive.

Case 36B. Competing reading and explicitness threshold

Compare the unsafe target, “The many responses proved the result certain.”, with “Although many people responded, the result remained uncertain.”. They are not two equally faithful stylistic options under the stated brief. The former introduces a relationship or implication that needs independent justification; the latter retains the intended boundary. A high count can coexist with selection bias or measurement limits. Test the exact meaning of the linking expression, not merely the dictionary meanings of the nouns and verbs. Ask if the target asserts causation, expectation violation, permission, sequence, evidential proof, or exclusivity. If it does, locate the corresponding source commitment.

Where the real document is genuinely ambiguous, there are three defensible editorial routes. Preserve the ambiguity through natural target phrasing, seek a qualified source-side answer, or use an approved translator note that separates explanation from the translated claim. Do not silently invent a causal explanation to make the paragraph feel complete. The specific competing possibility here is: Coverage errors could remain even with many responses. That possibility may or may not be correct, but its continued logical availability shows why an overconfident connective can be dangerous.

Case 36C. Reader-effect and risk diagnostic

A relation error is not just a linguistic blemish; it changes what a reader is licensed to conclude. In this case: Readers may trust a survey more than its design warrants. Review the target as a person relying on the document rather than as an editor looking for elegant wording. Which action, inference, expectation or accusation becomes more likely after the bad version? What would the intended audience believe if the first source proposition were true but the inferred causal or concessive link were false?

A practical severity review should record the precise sentence, the changed inference, the intended reader, and the possible consequence. If the sentence affects safety, rights, access, technical diagnosis, consent or financial decisions, an unsupported connective deserves more scrutiny than an ordinary style improvement. This scenario’s genre is public research, and the source meaning brief is Response volume does not alone establish representativeness or confidence. That is the benchmark for both the accuracy review and the final plain-language edit.

Case 36D. Evidence and author-query gate

An efficient reviewer does not write ‘unclear’ and stop. Ask a question that can be answered with evidence: What sampling and weighting issues influenced uncertainty? Include the source passage and the two candidate target readings. An author can then confirm the relation, provide missing context or explicitly leave the ambiguity open. The record should distinguish an author-confirmed meaning from a translator’s plausible interpretation and from a machine-generated suggestion.

After the answer, update the relation map, not only the visible sentence. If similar segments occur elsewhere, check them for consistency without mechanically propagating the same connective. Context may legitimately make the same pair of vocabulary items express a different relationship in a later document. For this exercise, the brief already supplies the acceptable decision: Response volume does not alone establish representativeness or confidence. The next step is to verify that the chosen target preserves it and creates no new attribution.

Case 36E. Adversarial paraphrase and back-reading

Take the faithful candidate — “Although many people responded, the result remained uncertain.” — and paraphrase it into a simple reader belief. Now take the unsafe candidate — “The many responses proved the result certain.” — and do the same. Write down the extra proposition, presupposition or causal claim that appears in the second version. A high count can coexist with selection bias or measurement limits. This comparison is stronger than asking whether both sentences sound natural. A fluent sentence can be misleading, and a back-translation can repeat the same mistake if the reviewer already accepts its assumptions.

For an adversarial check, imagine a reader who accepts the two reported events but rejects the suggested link. Would that reader still be entitled to believe the source? Consider the possibility that coverage errors could remain even with many responses. If yes, the stronger reading needs explicit textual or authorial support. The test does not require mathematical proof. It disciplines translation by making the asserted relationship visible and by preventing an intuition from hiding inside a connective.

Case 36F. Transfer drill and acceptance criterion

Transfer the same reasoning to another language, medium or genre without carrying over English word order. Keep the propositional facts and the stated relation fixed, but choose a natural construction suitable for the target audience. The final acceptable solution does not have to use the English connector in “Although many people responded, the result remained uncertain.” literally. It does have to preserve response volume does not alone establish representativeness or confidence. It must also avoid the new belief introduced by “The many responses proved the result certain.”.

The review is passed when a target reader can identify the two facts, draw only the source-authorized inference, and answer the practical question raised by this public research text without adopting an unsupported explanation. The specific failure warning is: Readers may trust a survey more than its design warrants. Keep the approval record with this sentence’s context, the reviewer’s decision and any author response. If a future revision changes the source relation, rerun this check rather than treating the existing translation as automatically safe.

Case 37: Creative writing discovery — Literary prose

Constructed source meaning: The room was empty. A cup on the table was still warm.

Intended meaning supplied by the exercise brief: The author intentionally permits inference without identifying who left or when. Candidate that fails the brief: “Someone had just escaped, leaving a warm cup.” Defensible rendering under the brief: “The room was empty. A cup on the table was still warm.”

Case 37A. Proposition isolation: what is actually stated

In this literary prose example, begin with the two source propositions without merging them into a single story: The room was empty. A cup on the table was still warm. The first analytical task is to preserve each proposition and its evidential status. The passage is not a licence to invent a hidden third event, an extra agent or an unreported decision. The apparently vivid explanation destroys a deliberate mystery by adding a person and action. A translator who starts by choosing a smooth connective may overlook that the connective itself carries a claim. Start by marking A and B, their subjects and their temporal or logical scope; then describe the relationship separately in an interpretation note. The relevant relation in the approved training brief is: The author intentionally permits inference without identifying who left or when.

A useful minimality test is to ask whether each standalone sentence remains true if the proposed relation is false. The cup might have been warmed by another source. That alternative demonstrates why the words on the page and the link between them require separate auditing. Only the brief for this constructed case resolves the intended link. In an actual translation, identify what language, adjacent document context or author confirmation performs the equivalent role. Do not claim certainty merely because an explanation seems intuitive.

Case 37B. Competing reading and explicitness threshold

Compare the unsafe target, “Someone had just escaped, leaving a warm cup.”, with “The room was empty. A cup on the table was still warm.”. They are not two equally faithful stylistic options under the stated brief. The former introduces a relationship or implication that needs independent justification; the latter retains the intended boundary. The apparently vivid explanation destroys a deliberate mystery by adding a person and action. Test the exact meaning of the linking expression, not merely the dictionary meanings of the nouns and verbs. Ask if the target asserts causation, expectation violation, permission, sequence, evidential proof, or exclusivity. If it does, locate the corresponding source commitment.

Where the real document is genuinely ambiguous, there are three defensible editorial routes. Preserve the ambiguity through natural target phrasing, seek a qualified source-side answer, or use an approved translator note that separates explanation from the translated claim. Do not silently invent a causal explanation to make the paragraph feel complete. The specific competing possibility here is: The cup might have been warmed by another source. That possibility may or may not be correct, but its continued logical availability shows why an overconfident connective can be dangerous.

Case 37C. Reader-effect and risk diagnostic

A relation error is not just a linguistic blemish; it changes what a reader is licensed to conclude. In this case: Readers lose the freedom to interpret the scene. Review the target as a person relying on the document rather than as an editor looking for elegant wording. Which action, inference, expectation or accusation becomes more likely after the bad version? What would the intended audience believe if the first source proposition were true but the inferred causal or concessive link were false?

A practical severity review should record the precise sentence, the changed inference, the intended reader, and the possible consequence. If the sentence affects safety, rights, access, technical diagnosis, consent or financial decisions, an unsupported connective deserves more scrutiny than an ordinary style improvement. This scenario’s genre is literary prose, and the source meaning brief is The author intentionally permits inference without identifying who left or when. That is the benchmark for both the accuracy review and the final plain-language edit.

Case 37D. Evidence and author-query gate

An efficient reviewer does not write ‘unclear’ and stop. Ask a question that can be answered with evidence: Does a later chapter confirm who used the cup? Include the source passage and the two candidate target readings. An author can then confirm the relation, provide missing context or explicitly leave the ambiguity open. The record should distinguish an author-confirmed meaning from a translator’s plausible interpretation and from a machine-generated suggestion.

After the answer, update the relation map, not only the visible sentence. If similar segments occur elsewhere, check them for consistency without mechanically propagating the same connective. Context may legitimately make the same pair of vocabulary items express a different relationship in a later document. For this exercise, the brief already supplies the acceptable decision: The author intentionally permits inference without identifying who left or when. The next step is to verify that the chosen target preserves it and creates no new attribution.

Case 37E. Adversarial paraphrase and back-reading

Take the faithful candidate — “The room was empty. A cup on the table was still warm.” — and paraphrase it into a simple reader belief. Now take the unsafe candidate — “Someone had just escaped, leaving a warm cup.” — and do the same. Write down the extra proposition, presupposition or causal claim that appears in the second version. The apparently vivid explanation destroys a deliberate mystery by adding a person and action. This comparison is stronger than asking whether both sentences sound natural. A fluent sentence can be misleading, and a back-translation can repeat the same mistake if the reviewer already accepts its assumptions.

For an adversarial check, imagine a reader who accepts the two reported events but rejects the suggested link. Would that reader still be entitled to believe the source? Consider the possibility that the cup might have been warmed by another source. If yes, the stronger reading needs explicit textual or authorial support. The test does not require mathematical proof. It disciplines translation by making the asserted relationship visible and by preventing an intuition from hiding inside a connective.

Case 37F. Transfer drill and acceptance criterion

Transfer the same reasoning to another language, medium or genre without carrying over English word order. Keep the propositional facts and the stated relation fixed, but choose a natural construction suitable for the target audience. The final acceptable solution does not have to use the English connector in “The room was empty. A cup on the table was still warm.” literally. It does have to preserve the author intentionally permits inference without identifying who left or when. It must also avoid the new belief introduced by “Someone had just escaped, leaving a warm cup.”.

The review is passed when a target reader can identify the two facts, draw only the source-authorized inference, and answer the practical question raised by this literary prose text without adopting an unsupported explanation. The specific failure warning is: Readers lose the freedom to interpret the scene. Keep the approval record with this sentence’s context, the reviewer’s decision and any author response. If a future revision changes the source relation, rerun this check rather than treating the existing translation as automatically safe.

Case 38: Policy rule and exception — User agreement

Constructed source meaning: Users can download their own data. Some archived files require a separate request.

Intended meaning supplied by the exercise brief: The second sentence restricts one subset without removing the general permission. Candidate that fails the brief: “All archived files may be downloaded instantly.” Defensible rendering under the brief: “Users can download their own data, although some archived files require a separate request.”

Case 38A. Proposition isolation: what is actually stated

In this user agreement example, begin with the two source propositions without merging them into a single story: Users can download their own data. Some archived files require a separate request. The first analytical task is to preserve each proposition and its evidential status. The passage is not a licence to invent a hidden third event, an extra agent or an unreported decision. An unqualified permission becomes materially misleading without its exception. A translator who starts by choosing a smooth connective may overlook that the connective itself carries a claim. Start by marking A and B, their subjects and their temporal or logical scope; then describe the relationship separately in an interpretation note. The relevant relation in the approved training brief is: The second sentence restricts one subset without removing the general permission.

A useful minimality test is to ask whether each standalone sentence remains true if the proposed relation is false. Different retention classes may have different rules. That alternative demonstrates why the words on the page and the link between them require separate auditing. Only the brief for this constructed case resolves the intended link. In an actual translation, identify what language, adjacent document context or author confirmation performs the equivalent role. Do not claim certainty merely because an explanation seems intuitive.

Case 38B. Competing reading and explicitness threshold

Compare the unsafe target, “All archived files may be downloaded instantly.”, with “Users can download their own data, although some archived files require a separate request.”. They are not two equally faithful stylistic options under the stated brief. The former introduces a relationship or implication that needs independent justification; the latter retains the intended boundary. An unqualified permission becomes materially misleading without its exception. Test the exact meaning of the linking expression, not merely the dictionary meanings of the nouns and verbs. Ask if the target asserts causation, expectation violation, permission, sequence, evidential proof, or exclusivity. If it does, locate the corresponding source commitment.

Where the real document is genuinely ambiguous, there are three defensible editorial routes. Preserve the ambiguity through natural target phrasing, seek a qualified source-side answer, or use an approved translator note that separates explanation from the translated claim. Do not silently invent a causal explanation to make the paragraph feel complete. The specific competing possibility here is: Different retention classes may have different rules. That possibility may or may not be correct, but its continued logical availability shows why an overconfident connective can be dangerous.

Case 38C. Reader-effect and risk diagnostic

A relation error is not just a linguistic blemish; it changes what a reader is licensed to conclude. In this case: A user may expect an unavailable feature or miss a request procedure. Review the target as a person relying on the document rather than as an editor looking for elegant wording. Which action, inference, expectation or accusation becomes more likely after the bad version? What would the intended audience believe if the first source proposition were true but the inferred causal or concessive link were false?

A practical severity review should record the precise sentence, the changed inference, the intended reader, and the possible consequence. If the sentence affects safety, rights, access, technical diagnosis, consent or financial decisions, an unsupported connective deserves more scrutiny than an ordinary style improvement. This scenario’s genre is user agreement, and the source meaning brief is The second sentence restricts one subset without removing the general permission. That is the benchmark for both the accuracy review and the final plain-language edit.

Case 38D. Evidence and author-query gate

An efficient reviewer does not write ‘unclear’ and stop. Ask a question that can be answered with evidence: Which archived file types trigger the separate workflow? Include the source passage and the two candidate target readings. An author can then confirm the relation, provide missing context or explicitly leave the ambiguity open. The record should distinguish an author-confirmed meaning from a translator’s plausible interpretation and from a machine-generated suggestion.

After the answer, update the relation map, not only the visible sentence. If similar segments occur elsewhere, check them for consistency without mechanically propagating the same connective. Context may legitimately make the same pair of vocabulary items express a different relationship in a later document. For this exercise, the brief already supplies the acceptable decision: The second sentence restricts one subset without removing the general permission. The next step is to verify that the chosen target preserves it and creates no new attribution.

Case 38E. Adversarial paraphrase and back-reading

Take the faithful candidate — “Users can download their own data, although some archived files require a separate request.” — and paraphrase it into a simple reader belief. Now take the unsafe candidate — “All archived files may be downloaded instantly.” — and do the same. Write down the extra proposition, presupposition or causal claim that appears in the second version. An unqualified permission becomes materially misleading without its exception. This comparison is stronger than asking whether both sentences sound natural. A fluent sentence can be misleading, and a back-translation can repeat the same mistake if the reviewer already accepts its assumptions.

For an adversarial check, imagine a reader who accepts the two reported events but rejects the suggested link. Would that reader still be entitled to believe the source? Consider the possibility that different retention classes may have different rules. If yes, the stronger reading needs explicit textual or authorial support. The test does not require mathematical proof. It disciplines translation by making the asserted relationship visible and by preventing an intuition from hiding inside a connective.

Case 38F. Transfer drill and acceptance criterion

Transfer the same reasoning to another language, medium or genre without carrying over English word order. Keep the propositional facts and the stated relation fixed, but choose a natural construction suitable for the target audience. The final acceptable solution does not have to use the English connector in “Users can download their own data, although some archived files require a separate request.” literally. It does have to preserve the second sentence restricts one subset without removing the general permission. It must also avoid the new belief introduced by “All archived files may be downloaded instantly.”.

The review is passed when a target reader can identify the two facts, draw only the source-authorized inference, and answer the practical question raised by this user agreement text without adopting an unsupported explanation. The specific failure warning is: A user may expect an unavailable feature or miss a request procedure. Keep the approval record with this sentence’s context, the reviewer’s decision and any author response. If a future revision changes the source relation, rerun this check rather than treating the existing translation as automatically safe.

Case 39: Community event and volunteer help — Organizer message

Constructed source meaning: The workshop filled quickly. Volunteers prepared additional chairs.

Intended meaning supplied by the exercise brief: The preparation responded to demand; no claim that capacity was formally increased. Candidate that fails the brief: “The venue increased its legal occupancy capacity.” Defensible rendering under the brief: “The workshop filled quickly, and volunteers prepared additional chairs.”

Case 39A. Proposition isolation: what is actually stated

In this organizer message example, begin with the two source propositions without merging them into a single story: The workshop filled quickly. Volunteers prepared additional chairs. The first analytical task is to preserve each proposition and its evidential status. The passage is not a licence to invent a hidden third event, an extra agent or an unreported decision. Extra chairs are not equivalent to a larger approved occupancy limit. A translator who starts by choosing a smooth connective may overlook that the connective itself carries a claim. Start by marking A and B, their subjects and their temporal or logical scope; then describe the relationship separately in an interpretation note. The relevant relation in the approved training brief is: The preparation responded to demand; no claim that capacity was formally increased.

A useful minimality test is to ask whether each standalone sentence remains true if the proposed relation is false. Fire-code capacity could remain unchanged. That alternative demonstrates why the words on the page and the link between them require separate auditing. Only the brief for this constructed case resolves the intended link. In an actual translation, identify what language, adjacent document context or author confirmation performs the equivalent role. Do not claim certainty merely because an explanation seems intuitive.

Case 39B. Competing reading and explicitness threshold

Compare the unsafe target, “The venue increased its legal occupancy capacity.”, with “The workshop filled quickly, and volunteers prepared additional chairs.”. They are not two equally faithful stylistic options under the stated brief. The former introduces a relationship or implication that needs independent justification; the latter retains the intended boundary. Extra chairs are not equivalent to a larger approved occupancy limit. Test the exact meaning of the linking expression, not merely the dictionary meanings of the nouns and verbs. Ask if the target asserts causation, expectation violation, permission, sequence, evidential proof, or exclusivity. If it does, locate the corresponding source commitment.

Where the real document is genuinely ambiguous, there are three defensible editorial routes. Preserve the ambiguity through natural target phrasing, seek a qualified source-side answer, or use an approved translator note that separates explanation from the translated claim. Do not silently invent a causal explanation to make the paragraph feel complete. The specific competing possibility here is: Fire-code capacity could remain unchanged. That possibility may or may not be correct, but its continued logical availability shows why an overconfident connective can be dangerous.

Case 39C. Reader-effect and risk diagnostic

A relation error is not just a linguistic blemish; it changes what a reader is licensed to conclude. In this case: Attendees may wrongly believe extra places are available. Review the target as a person relying on the document rather than as an editor looking for elegant wording. Which action, inference, expectation or accusation becomes more likely after the bad version? What would the intended audience believe if the first source proposition were true but the inferred causal or concessive link were false?

A practical severity review should record the precise sentence, the changed inference, the intended reader, and the possible consequence. If the sentence affects safety, rights, access, technical diagnosis, consent or financial decisions, an unsupported connective deserves more scrutiny than an ordinary style improvement. This scenario’s genre is organizer message, and the source meaning brief is The preparation responded to demand; no claim that capacity was formally increased. That is the benchmark for both the accuracy review and the final plain-language edit.

Case 39D. Evidence and author-query gate

An efficient reviewer does not write ‘unclear’ and stop. Ask a question that can be answered with evidence: Were registrations increased or was seating merely rearranged? Include the source passage and the two candidate target readings. An author can then confirm the relation, provide missing context or explicitly leave the ambiguity open. The record should distinguish an author-confirmed meaning from a translator’s plausible interpretation and from a machine-generated suggestion.

After the answer, update the relation map, not only the visible sentence. If similar segments occur elsewhere, check them for consistency without mechanically propagating the same connective. Context may legitimately make the same pair of vocabulary items express a different relationship in a later document. For this exercise, the brief already supplies the acceptable decision: The preparation responded to demand; no claim that capacity was formally increased. The next step is to verify that the chosen target preserves it and creates no new attribution.

Case 39E. Adversarial paraphrase and back-reading

Take the faithful candidate — “The workshop filled quickly, and volunteers prepared additional chairs.” — and paraphrase it into a simple reader belief. Now take the unsafe candidate — “The venue increased its legal occupancy capacity.” — and do the same. Write down the extra proposition, presupposition or causal claim that appears in the second version. Extra chairs are not equivalent to a larger approved occupancy limit. This comparison is stronger than asking whether both sentences sound natural. A fluent sentence can be misleading, and a back-translation can repeat the same mistake if the reviewer already accepts its assumptions.

For an adversarial check, imagine a reader who accepts the two reported events but rejects the suggested link. Would that reader still be entitled to believe the source? Consider the possibility that fire-code capacity could remain unchanged. If yes, the stronger reading needs explicit textual or authorial support. The test does not require mathematical proof. It disciplines translation by making the asserted relationship visible and by preventing an intuition from hiding inside a connective.

Case 39F. Transfer drill and acceptance criterion

Transfer the same reasoning to another language, medium or genre without carrying over English word order. Keep the propositional facts and the stated relation fixed, but choose a natural construction suitable for the target audience. The final acceptable solution does not have to use the English connector in “The workshop filled quickly, and volunteers prepared additional chairs.” literally. It does have to preserve the preparation responded to demand; no claim that capacity was formally increased. It must also avoid the new belief introduced by “The venue increased its legal occupancy capacity.”.

The review is passed when a target reader can identify the two facts, draw only the source-authorized inference, and answer the practical question raised by this organizer message text without adopting an unsupported explanation. The specific failure warning is: Attendees may wrongly believe extra places are available. Keep the approval record with this sentence’s context, the reviewer’s decision and any author response. If a future revision changes the source relation, rerun this check rather than treating the existing translation as automatically safe.

5. Failure-mode atlas: how fluent translations silently change reasoning

Failure 1: False cause

Symptom: A target ‘because’ turns a source sequence into a causal mechanism. Repair: Require explicit support from grammar, co-text or author evidence; otherwise separate the events. Reader consequence: A policy reader may treat an unproven explanation as an established rule. The diagnostic distinction matters because a polished target can preserve all content words while reshaping the reasoning between them. Highlight the exact source span and target span, state the extra or missing commitment, and decide whether the issue is minor, consequential or blocking for the intended use. Do not mark the entire paragraph wrong merely because it uses a different connective from a literal gloss; judge the inference it creates.

Quality-control exercise: write two alternative translations of this failure case, one with an overt relation marker and one with less overt marking. For each version, explain what a literal reader would be entitled to infer. Check whether the relation in false cause changes the status of evidence, action, expectation or speaker responsibility. The preferred version must be justified by the source, not by familiarity with English linking words. If the document supplies insufficient information, log a contextual query rather than installing a confident but unsupported answer.

Failure 2: False consequence

Symptom: A target ‘therefore’ claims that the second proposition follows from the first. Repair: Test whether reasonable counterexamples remain possible under the source. Reader consequence: A researcher may overstate the strength of an observed association. The diagnostic distinction matters because a polished target can preserve all content words while reshaping the reasoning between them. Highlight the exact source span and target span, state the extra or missing commitment, and decide whether the issue is minor, consequential or blocking for the intended use. Do not mark the entire paragraph wrong merely because it uses a different connective from a literal gloss; judge the inference it creates.

Quality-control exercise: write two alternative translations of this failure case, one with an overt relation marker and one with less overt marking. For each version, explain what a literal reader would be entitled to infer. Check whether the relation in false consequence changes the status of evidence, action, expectation or speaker responsibility. The preferred version must be justified by the source, not by familiarity with English linking words. If the document supplies insufficient information, log a contextual query rather than installing a confident but unsupported answer.

Failure 3: False contrast

Symptom: However suggests a competing dimension where the writer simply adds information. Repair: Ask whether a specific comparison dimension is intended. Reader consequence: A reader may assign controversy or opposition that did not exist. The diagnostic distinction matters because a polished target can preserve all content words while reshaping the reasoning between them. Highlight the exact source span and target span, state the extra or missing commitment, and decide whether the issue is minor, consequential or blocking for the intended use. Do not mark the entire paragraph wrong merely because it uses a different connective from a literal gloss; judge the inference it creates.

Quality-control exercise: write two alternative translations of this failure case, one with an overt relation marker and one with less overt marking. For each version, explain what a literal reader would be entitled to infer. Check whether the relation in false contrast changes the status of evidence, action, expectation or speaker responsibility. The preferred version must be justified by the source, not by familiarity with English linking words. If the document supplies insufficient information, log a contextual query rather than installing a confident but unsupported answer.

Failure 4: False concession

Symptom: Although creates a defeated expectation absent from the source. Repair: Identify the expectation explicitly before using a concessive form. Reader consequence: A biographical text may imply a person’s efforts were surprisingly futile. The diagnostic distinction matters because a polished target can preserve all content words while reshaping the reasoning between them. Highlight the exact source span and target span, state the extra or missing commitment, and decide whether the issue is minor, consequential or blocking for the intended use. Do not mark the entire paragraph wrong merely because it uses a different connective from a literal gloss; judge the inference it creates.

Quality-control exercise: write two alternative translations of this failure case, one with an overt relation marker and one with less overt marking. For each version, explain what a literal reader would be entitled to infer. Check whether the relation in false concession changes the status of evidence, action, expectation or speaker responsibility. The preferred version must be justified by the source, not by familiarity with English linking words. If the document supplies insufficient information, log a contextual query rather than installing a confident but unsupported answer.

Failure 5: Missing concession

Symptom: A bare and removes a deliberately paradoxical relation. Repair: Look for yet, nevertheless or equivalent source-side cues. Reader consequence: The author’s dramatic contrast can become a mundane list. The diagnostic distinction matters because a polished target can preserve all content words while reshaping the reasoning between them. Highlight the exact source span and target span, state the extra or missing commitment, and decide whether the issue is minor, consequential or blocking for the intended use. Do not mark the entire paragraph wrong merely because it uses a different connective from a literal gloss; judge the inference it creates.

Quality-control exercise: write two alternative translations of this failure case, one with an overt relation marker and one with less overt marking. For each version, explain what a literal reader would be entitled to infer. Check whether the relation in missing concession changes the status of evidence, action, expectation or speaker responsibility. The preferred version must be justified by the source, not by familiarity with English linking words. If the document supplies insufficient information, log a contextual query rather than installing a confident but unsupported answer.

Failure 6: Concession as cause

Symptom: A despite relationship is rewritten with because. Repair: Test the expectation direction using a two-column relation map. Reader consequence: The translation can reverse the point of an argument. The diagnostic distinction matters because a polished target can preserve all content words while reshaping the reasoning between them. Highlight the exact source span and target span, state the extra or missing commitment, and decide whether the issue is minor, consequential or blocking for the intended use. Do not mark the entire paragraph wrong merely because it uses a different connective from a literal gloss; judge the inference it creates.

Quality-control exercise: write two alternative translations of this failure case, one with an overt relation marker and one with less overt marking. For each version, explain what a literal reader would be entitled to infer. Check whether the relation in concession as cause changes the status of evidence, action, expectation or speaker responsibility. The preferred version must be justified by the source, not by familiarity with English linking words. If the document supplies insufficient information, log a contextual query rather than installing a confident but unsupported answer.

Failure 7: Causation from order

Symptom: Event order is treated as proof of influence. Repair: Apply the independence counterfactual and preserve after where appropriate. Reader consequence: Historical records can acquire invented causal claims. The diagnostic distinction matters because a polished target can preserve all content words while reshaping the reasoning between them. Highlight the exact source span and target span, state the extra or missing commitment, and decide whether the issue is minor, consequential or blocking for the intended use. Do not mark the entire paragraph wrong merely because it uses a different connective from a literal gloss; judge the inference it creates.

Quality-control exercise: write two alternative translations of this failure case, one with an overt relation marker and one with less overt marking. For each version, explain what a literal reader would be entitled to infer. Check whether the relation in causation from order changes the status of evidence, action, expectation or speaker responsibility. The preferred version must be justified by the source, not by familiarity with English linking words. If the document supplies insufficient information, log a contextual query rather than installing a confident but unsupported answer.

Failure 8: Condition to prediction

Symptom: If becomes when, presupposing a triggering event will occur. Repair: Check whether the source guarantees, merely allows or does not anticipate the trigger. Reader consequence: Applicants can rely on an outcome never promised. The diagnostic distinction matters because a polished target can preserve all content words while reshaping the reasoning between them. Highlight the exact source span and target span, state the extra or missing commitment, and decide whether the issue is minor, consequential or blocking for the intended use. Do not mark the entire paragraph wrong merely because it uses a different connective from a literal gloss; judge the inference it creates.

Quality-control exercise: write two alternative translations of this failure case, one with an overt relation marker and one with less overt marking. For each version, explain what a literal reader would be entitled to infer. Check whether the relation in condition to prediction changes the status of evidence, action, expectation or speaker responsibility. The preferred version must be justified by the source, not by familiarity with English linking words. If the document supplies insufficient information, log a contextual query rather than installing a confident but unsupported answer.

Failure 9: Prediction to condition

Symptom: Will is softened to if without authorization. Repair: Preserve the original level of commitment and actual triggering conditions. Reader consequence: A firm undertaking can appear discretionary. The diagnostic distinction matters because a polished target can preserve all content words while reshaping the reasoning between them. Highlight the exact source span and target span, state the extra or missing commitment, and decide whether the issue is minor, consequential or blocking for the intended use. Do not mark the entire paragraph wrong merely because it uses a different connective from a literal gloss; judge the inference it creates.

Quality-control exercise: write two alternative translations of this failure case, one with an overt relation marker and one with less overt marking. For each version, explain what a literal reader would be entitled to infer. Check whether the relation in prediction to condition changes the status of evidence, action, expectation or speaker responsibility. The preferred version must be justified by the source, not by familiarity with English linking words. If the document supplies insufficient information, log a contextual query rather than installing a confident but unsupported answer.

Failure 10: Exclusive list

Symptom: For example becomes namely or only. Repair: Distinguish illustrations from exhaustive membership. Reader consequence: Readers may mistakenly exclude valid options. The diagnostic distinction matters because a polished target can preserve all content words while reshaping the reasoning between them. Highlight the exact source span and target span, state the extra or missing commitment, and decide whether the issue is minor, consequential or blocking for the intended use. Do not mark the entire paragraph wrong merely because it uses a different connective from a literal gloss; judge the inference it creates.

Quality-control exercise: write two alternative translations of this failure case, one with an overt relation marker and one with less overt marking. For each version, explain what a literal reader would be entitled to infer. Check whether the relation in exclusive list changes the status of evidence, action, expectation or speaker responsibility. The preferred version must be justified by the source, not by familiarity with English linking words. If the document supplies insufficient information, log a contextual query rather than installing a confident but unsupported answer.

Failure 11: Exhaustive list expanded

Symptom: A closed set becomes such as plus an open-ended continuation. Repair: Check list-introducing phrases and any defined terms. Reader consequence: A policy may acquire unauthorized options. The diagnostic distinction matters because a polished target can preserve all content words while reshaping the reasoning between them. Highlight the exact source span and target span, state the extra or missing commitment, and decide whether the issue is minor, consequential or blocking for the intended use. Do not mark the entire paragraph wrong merely because it uses a different connective from a literal gloss; judge the inference it creates.

Quality-control exercise: write two alternative translations of this failure case, one with an overt relation marker and one with less overt marking. For each version, explain what a literal reader would be entitled to infer. Check whether the relation in exhaustive list expanded changes the status of evidence, action, expectation or speaker responsibility. The preferred version must be justified by the source, not by familiarity with English linking words. If the document supplies insufficient information, log a contextual query rather than installing a confident but unsupported answer.

Failure 12: Background promoted to proof

Symptom: Historical context becomes the stated reason for a current decision. Repair: Check whether the source assigns explanatory force to the background. Reader consequence: Readers may mistake context for documented evidence. The diagnostic distinction matters because a polished target can preserve all content words while reshaping the reasoning between them. Highlight the exact source span and target span, state the extra or missing commitment, and decide whether the issue is minor, consequential or blocking for the intended use. Do not mark the entire paragraph wrong merely because it uses a different connective from a literal gloss; judge the inference it creates.

Quality-control exercise: write two alternative translations of this failure case, one with an overt relation marker and one with less overt marking. For each version, explain what a literal reader would be entitled to infer. Check whether the relation in background promoted to proof changes the status of evidence, action, expectation or speaker responsibility. The preferred version must be justified by the source, not by familiarity with English linking words. If the document supplies insufficient information, log a contextual query rather than installing a confident but unsupported answer.

Failure 13: Explanation promoted to endorsement

Symptom: A quoted reason is presented as narrator-certified fact. Repair: Carry speaker attribution through the whole relation. Reader consequence: Journalistic independence can be lost. The diagnostic distinction matters because a polished target can preserve all content words while reshaping the reasoning between them. Highlight the exact source span and target span, state the extra or missing commitment, and decide whether the issue is minor, consequential or blocking for the intended use. Do not mark the entire paragraph wrong merely because it uses a different connective from a literal gloss; judge the inference it creates.

Quality-control exercise: write two alternative translations of this failure case, one with an overt relation marker and one with less overt marking. For each version, explain what a literal reader would be entitled to infer. Check whether the relation in explanation promoted to endorsement changes the status of evidence, action, expectation or speaker responsibility. The preferred version must be justified by the source, not by familiarity with English linking words. If the document supplies insufficient information, log a contextual query rather than installing a confident but unsupported answer.

Failure 14: Missing attribution

Symptom: The phrase according to is dropped when the causal claim is repeated. Repair: Keep the attribution boundary across sentences. Reader consequence: A disputed allegation can become an unqualified claim. The diagnostic distinction matters because a polished target can preserve all content words while reshaping the reasoning between them. Highlight the exact source span and target span, state the extra or missing commitment, and decide whether the issue is minor, consequential or blocking for the intended use. Do not mark the entire paragraph wrong merely because it uses a different connective from a literal gloss; judge the inference it creates.

Quality-control exercise: write two alternative translations of this failure case, one with an overt relation marker and one with less overt marking. For each version, explain what a literal reader would be entitled to infer. Check whether the relation in missing attribution changes the status of evidence, action, expectation or speaker responsibility. The preferred version must be justified by the source, not by familiarity with English linking words. If the document supplies insufficient information, log a contextual query rather than installing a confident but unsupported answer.

Failure 15: Wrong direction

Symptom: B explains A becomes A explains B. Repair: Make arrows between spans, not just a bag of relation words. Reader consequence: A procedural sequence can become illogical. The diagnostic distinction matters because a polished target can preserve all content words while reshaping the reasoning between them. Highlight the exact source span and target span, state the extra or missing commitment, and decide whether the issue is minor, consequential or blocking for the intended use. Do not mark the entire paragraph wrong merely because it uses a different connective from a literal gloss; judge the inference it creates.

Quality-control exercise: write two alternative translations of this failure case, one with an overt relation marker and one with less overt marking. For each version, explain what a literal reader would be entitled to infer. Check whether the relation in wrong direction changes the status of evidence, action, expectation or speaker responsibility. The preferred version must be justified by the source, not by familiarity with English linking words. If the document supplies insufficient information, log a contextual query rather than installing a confident but unsupported answer.

Failure 16: Wrong scope

Symptom: However is attached to the nearest sentence instead of the previous argument. Repair: Annotate span boundaries at paragraph level. Reader consequence: A local exception can wrongly negate the main conclusion. The diagnostic distinction matters because a polished target can preserve all content words while reshaping the reasoning between them. Highlight the exact source span and target span, state the extra or missing commitment, and decide whether the issue is minor, consequential or blocking for the intended use. Do not mark the entire paragraph wrong merely because it uses a different connective from a literal gloss; judge the inference it creates.

Quality-control exercise: write two alternative translations of this failure case, one with an overt relation marker and one with less overt marking. For each version, explain what a literal reader would be entitled to infer. Check whether the relation in wrong scope changes the status of evidence, action, expectation or speaker responsibility. The preferred version must be justified by the source, not by familiarity with English linking words. If the document supplies insufficient information, log a contextual query rather than installing a confident but unsupported answer.

Failure 17: Flattened hierarchy

Symptom: Several nested conditions are converted to a flat chain. Repair: Draw brackets around governed clauses before editing. Reader consequence: Instructions may authorize action under the wrong condition. The diagnostic distinction matters because a polished target can preserve all content words while reshaping the reasoning between them. Highlight the exact source span and target span, state the extra or missing commitment, and decide whether the issue is minor, consequential or blocking for the intended use. Do not mark the entire paragraph wrong merely because it uses a different connective from a literal gloss; judge the inference it creates.

Quality-control exercise: write two alternative translations of this failure case, one with an overt relation marker and one with less overt marking. For each version, explain what a literal reader would be entitled to infer. Check whether the relation in flattened hierarchy changes the status of evidence, action, expectation or speaker responsibility. The preferred version must be justified by the source, not by familiarity with English linking words. If the document supplies insufficient information, log a contextual query rather than installing a confident but unsupported answer.

Failure 18: Inverted exception

Symptom: Unless is confused with if or only if. Repair: Write a plain-language truth table for triggering situations. Reader consequence: A user may do the opposite of the source requirement. The diagnostic distinction matters because a polished target can preserve all content words while reshaping the reasoning between them. Highlight the exact source span and target span, state the extra or missing commitment, and decide whether the issue is minor, consequential or blocking for the intended use. Do not mark the entire paragraph wrong merely because it uses a different connective from a literal gloss; judge the inference it creates.

Quality-control exercise: write two alternative translations of this failure case, one with an overt relation marker and one with less overt marking. For each version, explain what a literal reader would be entitled to infer. Check whether the relation in inverted exception changes the status of evidence, action, expectation or speaker responsibility. The preferred version must be justified by the source, not by familiarity with English linking words. If the document supplies insufficient information, log a contextual query rather than installing a confident but unsupported answer.

Failure 19: Temporal while misread

Symptom: Simultaneity is translated as concession, or the reverse. Repair: Ask whether both clauses describe overlapping events or contrastive propositions. Reader consequence: Maintenance steps may become ill-timed. The diagnostic distinction matters because a polished target can preserve all content words while reshaping the reasoning between them. Highlight the exact source span and target span, state the extra or missing commitment, and decide whether the issue is minor, consequential or blocking for the intended use. Do not mark the entire paragraph wrong merely because it uses a different connective from a literal gloss; judge the inference it creates.

Quality-control exercise: write two alternative translations of this failure case, one with an overt relation marker and one with less overt marking. For each version, explain what a literal reader would be entitled to infer. Check whether the relation in temporal while misread changes the status of evidence, action, expectation or speaker responsibility. The preferred version must be justified by the source, not by familiarity with English linking words. If the document supplies insufficient information, log a contextual query rather than installing a confident but unsupported answer.

Failure 20: Since ambiguity ignored

Symptom: Since is treated as temporal though the sentence gives a reason. Repair: Test duration phrases and causal paraphrases. Reader consequence: A period of time may be invented or lost. The diagnostic distinction matters because a polished target can preserve all content words while reshaping the reasoning between them. Highlight the exact source span and target span, state the extra or missing commitment, and decide whether the issue is minor, consequential or blocking for the intended use. Do not mark the entire paragraph wrong merely because it uses a different connective from a literal gloss; judge the inference it creates.

Quality-control exercise: write two alternative translations of this failure case, one with an overt relation marker and one with less overt marking. For each version, explain what a literal reader would be entitled to infer. Check whether the relation in since ambiguity ignored changes the status of evidence, action, expectation or speaker responsibility. The preferred version must be justified by the source, not by familiarity with English linking words. If the document supplies insufficient information, log a contextual query rather than installing a confident but unsupported answer.

Failure 21: Purpose mistaken for consequence

Symptom: So that expresses an intended goal but is rewritten as a realized result. Repair: Distinguish intention from achieved outcome. Reader consequence: A performance claim may become falsely guaranteed. The diagnostic distinction matters because a polished target can preserve all content words while reshaping the reasoning between them. Highlight the exact source span and target span, state the extra or missing commitment, and decide whether the issue is minor, consequential or blocking for the intended use. Do not mark the entire paragraph wrong merely because it uses a different connective from a literal gloss; judge the inference it creates.

Quality-control exercise: write two alternative translations of this failure case, one with an overt relation marker and one with less overt marking. For each version, explain what a literal reader would be entitled to infer. Check whether the relation in purpose mistaken for consequence changes the status of evidence, action, expectation or speaker responsibility. The preferred version must be justified by the source, not by familiarity with English linking words. If the document supplies insufficient information, log a contextual query rather than installing a confident but unsupported answer.

Failure 22: Reason mistaken for evidence

Symptom: The cause of an event becomes evidence for a general conclusion. Repair: Identify whether the span explains an event or supports belief. Reader consequence: A research discussion can overstate the claim’s support. The diagnostic distinction matters because a polished target can preserve all content words while reshaping the reasoning between them. Highlight the exact source span and target span, state the extra or missing commitment, and decide whether the issue is minor, consequential or blocking for the intended use. Do not mark the entire paragraph wrong merely because it uses a different connective from a literal gloss; judge the inference it creates.

Quality-control exercise: write two alternative translations of this failure case, one with an overt relation marker and one with less overt marking. For each version, explain what a literal reader would be entitled to infer. Check whether the relation in reason mistaken for evidence changes the status of evidence, action, expectation or speaker responsibility. The preferred version must be justified by the source, not by familiarity with English linking words. If the document supplies insufficient information, log a contextual query rather than installing a confident but unsupported answer.

Failure 23: Evidence mistaken for cause

Symptom: Corroborating evidence is cast as an event’s physical cause. Repair: Keep evidential and real-world dependencies separate. Reader consequence: An investigation may assign wrong mechanisms. The diagnostic distinction matters because a polished target can preserve all content words while reshaping the reasoning between them. Highlight the exact source span and target span, state the extra or missing commitment, and decide whether the issue is minor, consequential or blocking for the intended use. Do not mark the entire paragraph wrong merely because it uses a different connective from a literal gloss; judge the inference it creates.

Quality-control exercise: write two alternative translations of this failure case, one with an overt relation marker and one with less overt marking. For each version, explain what a literal reader would be entitled to infer. Check whether the relation in evidence mistaken for cause changes the status of evidence, action, expectation or speaker responsibility. The preferred version must be justified by the source, not by familiarity with English linking words. If the document supplies insufficient information, log a contextual query rather than installing a confident but unsupported answer.

Failure 24: Conjunction erased

Symptom: Neither A nor B is reduced to an alternative relation. Repair: Check polarity over the entire coordinated structure. Reader consequence: A safety record may claim one alert fired when neither did. The diagnostic distinction matters because a polished target can preserve all content words while reshaping the reasoning between them. Highlight the exact source span and target span, state the extra or missing commitment, and decide whether the issue is minor, consequential or blocking for the intended use. Do not mark the entire paragraph wrong merely because it uses a different connective from a literal gloss; judge the inference it creates.

Quality-control exercise: write two alternative translations of this failure case, one with an overt relation marker and one with less overt marking. For each version, explain what a literal reader would be entitled to infer. Check whether the relation in conjunction erased changes the status of evidence, action, expectation or speaker responsibility. The preferred version must be justified by the source, not by familiarity with English linking words. If the document supplies insufficient information, log a contextual query rather than installing a confident but unsupported answer.

Failure 25: Contrast without symmetry

Symptom: Whereas pairs comparable dimensions but translation changes one dimension. Repair: Align comparable attributes on both sides. Reader consequence: Product comparisons can become incoherent. The diagnostic distinction matters because a polished target can preserve all content words while reshaping the reasoning between them. Highlight the exact source span and target span, state the extra or missing commitment, and decide whether the issue is minor, consequential or blocking for the intended use. Do not mark the entire paragraph wrong merely because it uses a different connective from a literal gloss; judge the inference it creates.

Quality-control exercise: write two alternative translations of this failure case, one with an overt relation marker and one with less overt marking. For each version, explain what a literal reader would be entitled to infer. Check whether the relation in contrast without symmetry changes the status of evidence, action, expectation or speaker responsibility. The preferred version must be justified by the source, not by familiarity with English linking words. If the document supplies insufficient information, log a contextual query rather than installing a confident but unsupported answer.

Failure 26: Restatement with new conditions

Symptom: In other words hides an extra requirement. Repair: Compare paraphrases for added truth conditions. Reader consequence: A plain-language notice may grow stricter than the source. The diagnostic distinction matters because a polished target can preserve all content words while reshaping the reasoning between them. Highlight the exact source span and target span, state the extra or missing commitment, and decide whether the issue is minor, consequential or blocking for the intended use. Do not mark the entire paragraph wrong merely because it uses a different connective from a literal gloss; judge the inference it creates.

Quality-control exercise: write two alternative translations of this failure case, one with an overt relation marker and one with less overt marking. For each version, explain what a literal reader would be entitled to infer. Check whether the relation in restatement with new conditions changes the status of evidence, action, expectation or speaker responsibility. The preferred version must be justified by the source, not by familiarity with English linking words. If the document supplies insufficient information, log a contextual query rather than installing a confident but unsupported answer.

Failure 27: Topic transition mistaken for inference

Symptom: Anyway or now is treated like therefore. Repair: Check whether the marker guides conversation rather than logic. Reader consequence: A speaker’s change of subject can be misread as a conclusion. The diagnostic distinction matters because a polished target can preserve all content words while reshaping the reasoning between them. Highlight the exact source span and target span, state the extra or missing commitment, and decide whether the issue is minor, consequential or blocking for the intended use. Do not mark the entire paragraph wrong merely because it uses a different connective from a literal gloss; judge the inference it creates.

Quality-control exercise: write two alternative translations of this failure case, one with an overt relation marker and one with less overt marking. For each version, explain what a literal reader would be entitled to infer. Check whether the relation in topic transition mistaken for inference changes the status of evidence, action, expectation or speaker responsibility. The preferred version must be justified by the source, not by familiarity with English linking words. If the document supplies insufficient information, log a contextual query rather than installing a confident but unsupported answer.

Failure 28: Over-normalized paragraph

Symptom: An editor adds linkers to every sentence for smoothness. Repair: Review each addition against the source relation map. Reader consequence: An exploratory text can turn into a falsely certain argument. The diagnostic distinction matters because a polished target can preserve all content words while reshaping the reasoning between them. Highlight the exact source span and target span, state the extra or missing commitment, and decide whether the issue is minor, consequential or blocking for the intended use. Do not mark the entire paragraph wrong merely because it uses a different connective from a literal gloss; judge the inference it creates.

Quality-control exercise: write two alternative translations of this failure case, one with an overt relation marker and one with less overt marking. For each version, explain what a literal reader would be entitled to infer. Check whether the relation in over-normalized paragraph changes the status of evidence, action, expectation or speaker responsibility. The preferred version must be justified by the source, not by familiarity with English linking words. If the document supplies insufficient information, log a contextual query rather than installing a confident but unsupported answer.

Failure 29: Under-specified target

Symptom: Removing all linkers makes a clear relation obscure to the target reader. Repair: Retain enough explicit marking for functional comprehension. Reader consequence: Readers can misorder instructions or miss an exception. The diagnostic distinction matters because a polished target can preserve all content words while reshaping the reasoning between them. Highlight the exact source span and target span, state the extra or missing commitment, and decide whether the issue is minor, consequential or blocking for the intended use. Do not mark the entire paragraph wrong merely because it uses a different connective from a literal gloss; judge the inference it creates.

Quality-control exercise: write two alternative translations of this failure case, one with an overt relation marker and one with less overt marking. For each version, explain what a literal reader would be entitled to infer. Check whether the relation in under-specified target changes the status of evidence, action, expectation or speaker responsibility. The preferred version must be justified by the source, not by familiarity with English linking words. If the document supplies insufficient information, log a contextual query rather than installing a confident but unsupported answer.

Failure 30: Untracked source revision

Symptom: A causal explanation changes, but the translation keeps an old connective. Repair: Track modified source spans and revalidate adjacent relations. Reader consequence: A corrected document can still distribute a superseded rationale. The diagnostic distinction matters because a polished target can preserve all content words while reshaping the reasoning between them. Highlight the exact source span and target span, state the extra or missing commitment, and decide whether the issue is minor, consequential or blocking for the intended use. Do not mark the entire paragraph wrong merely because it uses a different connective from a literal gloss; judge the inference it creates.

Quality-control exercise: write two alternative translations of this failure case, one with an overt relation marker and one with less overt marking. For each version, explain what a literal reader would be entitled to infer. Check whether the relation in untracked source revision changes the status of evidence, action, expectation or speaker responsibility. The preferred version must be justified by the source, not by familiarity with English linking words. If the document supplies insufficient information, log a contextual query rather than installing a confident but unsupported answer.

Failure 31: Machine output accepted for fluency

Symptom: A model supplies a plausible because, so or although without support. Repair: Run targeted source-alignment and contradiction checks, not only fluency review. Reader consequence: High-confidence language hides low-confidence reasoning. The diagnostic distinction matters because a polished target can preserve all content words while reshaping the reasoning between them. Highlight the exact source span and target span, state the extra or missing commitment, and decide whether the issue is minor, consequential or blocking for the intended use. Do not mark the entire paragraph wrong merely because it uses a different connective from a literal gloss; judge the inference it creates.

Quality-control exercise: write two alternative translations of this failure case, one with an overt relation marker and one with less overt marking. For each version, explain what a literal reader would be entitled to infer. Check whether the relation in machine output accepted for fluency changes the status of evidence, action, expectation or speaker responsibility. The preferred version must be justified by the source, not by familiarity with English linking words. If the document supplies insufficient information, log a contextual query rather than installing a confident but unsupported answer.

Failure 32: Segmented translation memory error

Symptom: A clause is translated without the paragraph that explains it. Repair: Include context windows or annotations for relation-critical segments. Reader consequence: A reused sentence acquires the wrong link in a new setting. The diagnostic distinction matters because a polished target can preserve all content words while reshaping the reasoning between them. Highlight the exact source span and target span, state the extra or missing commitment, and decide whether the issue is minor, consequential or blocking for the intended use. Do not mark the entire paragraph wrong merely because it uses a different connective from a literal gloss; judge the inference it creates.

Quality-control exercise: write two alternative translations of this failure case, one with an overt relation marker and one with less overt marking. For each version, explain what a literal reader would be entitled to infer. Check whether the relation in segmented translation memory error changes the status of evidence, action, expectation or speaker responsibility. The preferred version must be justified by the source, not by familiarity with English linking words. If the document supplies insufficient information, log a contextual query rather than installing a confident but unsupported answer.

Failure 33: Speaker-turn confusion

Symptom: A reply’s ‘yes, but’ is read as an agreement with the whole previous claim. Repair: Identify which proposition is accepted and which is challenged. Reader consequence: Minutes may record consensus where disagreement remains. The diagnostic distinction matters because a polished target can preserve all content words while reshaping the reasoning between them. Highlight the exact source span and target span, state the extra or missing commitment, and decide whether the issue is minor, consequential or blocking for the intended use. Do not mark the entire paragraph wrong merely because it uses a different connective from a literal gloss; judge the inference it creates.

Quality-control exercise: write two alternative translations of this failure case, one with an overt relation marker and one with less overt marking. For each version, explain what a literal reader would be entitled to infer. Check whether the relation in speaker-turn confusion changes the status of evidence, action, expectation or speaker responsibility. The preferred version must be justified by the source, not by familiarity with English linking words. If the document supplies insufficient information, log a contextual query rather than installing a confident but unsupported answer.

Failure 34: Headings treated as causes

Symptom: A section heading such as ‘Results’ is made a sentence-level causal assertion. Repair: Separate document navigation from propositional links. Reader consequence: A report can imply proof from mere organization. The diagnostic distinction matters because a polished target can preserve all content words while reshaping the reasoning between them. Highlight the exact source span and target span, state the extra or missing commitment, and decide whether the issue is minor, consequential or blocking for the intended use. Do not mark the entire paragraph wrong merely because it uses a different connective from a literal gloss; judge the inference it creates.

Quality-control exercise: write two alternative translations of this failure case, one with an overt relation marker and one with less overt marking. For each version, explain what a literal reader would be entitled to infer. Check whether the relation in headings treated as causes changes the status of evidence, action, expectation or speaker responsibility. The preferred version must be justified by the source, not by familiarity with English linking words. If the document supplies insufficient information, log a contextual query rather than installing a confident but unsupported answer.

Failure 35: Undocumented adaptation

Symptom: An editor makes an explanation explicit without recording authorization. Repair: Distinguish target-language accessibility adaptation from source claims. Reader consequence: Later editors may mistake a clarification for authoritative source content. The diagnostic distinction matters because a polished target can preserve all content words while reshaping the reasoning between them. Highlight the exact source span and target span, state the extra or missing commitment, and decide whether the issue is minor, consequential or blocking for the intended use. Do not mark the entire paragraph wrong merely because it uses a different connective from a literal gloss; judge the inference it creates.

Quality-control exercise: write two alternative translations of this failure case, one with an overt relation marker and one with less overt marking. For each version, explain what a literal reader would be entitled to infer. Check whether the relation in undocumented adaptation changes the status of evidence, action, expectation or speaker responsibility. The preferred version must be justified by the source, not by familiarity with English linking words. If the document supplies insufficient information, log a contextual query rather than installing a confident but unsupported answer.

Failure 36: One-size-fits-all connective

Symptom: A fixed glossary maps each marker to one target-language word. Repair: Select by function, scope, genre and register, not token identity. Reader consequence: Correct dictionaries can still produce incorrect text relations. The diagnostic distinction matters because a polished target can preserve all content words while reshaping the reasoning between them. Highlight the exact source span and target span, state the extra or missing commitment, and decide whether the issue is minor, consequential or blocking for the intended use. Do not mark the entire paragraph wrong merely because it uses a different connective from a literal gloss; judge the inference it creates.

Quality-control exercise: write two alternative translations of this failure case, one with an overt relation marker and one with less overt marking. For each version, explain what a literal reader would be entitled to infer. Check whether the relation in one-size-fits-all connective changes the status of evidence, action, expectation or speaker responsibility. The preferred version must be justified by the source, not by familiarity with English linking words. If the document supplies insufficient information, log a contextual query rather than installing a confident but unsupported answer.

6. Transfer workshops: practice with controlled relation changes

Workshop 1: Research abstract

Source-side meaning brief: A measured difference may reflect a mechanism, but the analysis has not established cause. First identify the two independent claims and the relation that the brief licenses. Separate observational data from causal interpretation and test whether the conclusion remains hedged. Draft a version with as little overt marking as possible while still being clear, then draft a version with the most specific connective you can justify. Explain why each candidate does or does not preserve the original level of commitment. The point of this exercise is the decision process, not memorizing a single answer string.

Adversarial extension: imagine a target reader under time pressure who sees only the translated paragraph. Write one false inference that might be encouraged by adding because, therefore, although, only if, while or for example in the wrong place. Then replace the linker or change sentence structure so that the source meaning remains intact. Check whether the second span is an observation, reason, exception, limitation, evidence point or independent fact. If the target becomes less fluent after correcting a relation, improve the sentence while keeping the repaired relationship stable.

Completion standard: another reviewer should be able to explain why your chosen relation is supported, point to the source evidence and identify which competing rendering was excluded. Separate observational data from causal interpretation and test whether the conclusion remains hedged. A source–target sheet should record the approved wording and an unresolved-author-query field. This enables later reuse without pretending that all instances of the same linking word across languages have the same discourse function. Revisit the exercise after reviewing other domains: a useful skill transfers even when the surface vocabulary changes completely.

Workshop 2: Weather advisory

Source-side meaning brief: Rain is forecast; organizers have not announced a cancellation. First identify the two independent claims and the relation that the brief licenses. Preserve forecast, uncertainty and decision status, and do not invent an operative conditional rule. Draft a version with as little overt marking as possible while still being clear, then draft a version with the most specific connective you can justify. Explain why each candidate does or does not preserve the original level of commitment. The point of this exercise is the decision process, not memorizing a single answer string.

Adversarial extension: imagine a target reader under time pressure who sees only the translated paragraph. Write one false inference that might be encouraged by adding because, therefore, although, only if, while or for example in the wrong place. Then replace the linker or change sentence structure so that the source meaning remains intact. Check whether the second span is an observation, reason, exception, limitation, evidence point or independent fact. If the target becomes less fluent after correcting a relation, improve the sentence while keeping the repaired relationship stable.

Completion standard: another reviewer should be able to explain why your chosen relation is supported, point to the source evidence and identify which competing rendering was excluded. Preserve forecast, uncertainty and decision status, and do not invent an operative conditional rule. A source–target sheet should record the approved wording and an unresolved-author-query field. This enables later reuse without pretending that all instances of the same linking word across languages have the same discourse function. Revisit the exercise after reviewing other domains: a useful skill transfers even when the surface vocabulary changes completely.

Workshop 3: Software interface

Source-side meaning brief: An icon is visible, but the feature is unavailable until setup completes. First identify the two independent claims and the relation that the brief licenses. Mark the condition governing access without implying setup has happened. Draft a version with as little overt marking as possible while still being clear, then draft a version with the most specific connective you can justify. Explain why each candidate does or does not preserve the original level of commitment. The point of this exercise is the decision process, not memorizing a single answer string.

Adversarial extension: imagine a target reader under time pressure who sees only the translated paragraph. Write one false inference that might be encouraged by adding because, therefore, although, only if, while or for example in the wrong place. Then replace the linker or change sentence structure so that the source meaning remains intact. Check whether the second span is an observation, reason, exception, limitation, evidence point or independent fact. If the target becomes less fluent after correcting a relation, improve the sentence while keeping the repaired relationship stable.

Completion standard: another reviewer should be able to explain why your chosen relation is supported, point to the source evidence and identify which competing rendering was excluded. Mark the condition governing access without implying setup has happened. A source–target sheet should record the approved wording and an unresolved-author-query field. This enables later reuse without pretending that all instances of the same linking word across languages have the same discourse function. Revisit the exercise after reviewing other domains: a useful skill transfers even when the surface vocabulary changes completely.

Workshop 4: Public consultation

Source-side meaning brief: Some respondents approve of the project; others ask for more evidence. First identify the two independent claims and the relation that the brief licenses. Present differing positions without presuming hostility or assigning bad motives. Draft a version with as little overt marking as possible while still being clear, then draft a version with the most specific connective you can justify. Explain why each candidate does or does not preserve the original level of commitment. The point of this exercise is the decision process, not memorizing a single answer string.

Adversarial extension: imagine a target reader under time pressure who sees only the translated paragraph. Write one false inference that might be encouraged by adding because, therefore, although, only if, while or for example in the wrong place. Then replace the linker or change sentence structure so that the source meaning remains intact. Check whether the second span is an observation, reason, exception, limitation, evidence point or independent fact. If the target becomes less fluent after correcting a relation, improve the sentence while keeping the repaired relationship stable.

Completion standard: another reviewer should be able to explain why your chosen relation is supported, point to the source evidence and identify which competing rendering was excluded. Present differing positions without presuming hostility or assigning bad motives. A source–target sheet should record the approved wording and an unresolved-author-query field. This enables later reuse without pretending that all instances of the same linking word across languages have the same discourse function. Revisit the exercise after reviewing other domains: a useful skill transfers even when the surface vocabulary changes completely.

Workshop 5: University policy

Source-side meaning brief: Students may appeal; an appeal does not guarantee a changed grade. First identify the two independent claims and the relation that the brief licenses. Distinguish a procedural right from an outcome entitlement. Draft a version with as little overt marking as possible while still being clear, then draft a version with the most specific connective you can justify. Explain why each candidate does or does not preserve the original level of commitment. The point of this exercise is the decision process, not memorizing a single answer string.

Adversarial extension: imagine a target reader under time pressure who sees only the translated paragraph. Write one false inference that might be encouraged by adding because, therefore, although, only if, while or for example in the wrong place. Then replace the linker or change sentence structure so that the source meaning remains intact. Check whether the second span is an observation, reason, exception, limitation, evidence point or independent fact. If the target becomes less fluent after correcting a relation, improve the sentence while keeping the repaired relationship stable.

Completion standard: another reviewer should be able to explain why your chosen relation is supported, point to the source evidence and identify which competing rendering was excluded. Distinguish a procedural right from an outcome entitlement. A source–target sheet should record the approved wording and an unresolved-author-query field. This enables later reuse without pretending that all instances of the same linking word across languages have the same discourse function. Revisit the exercise after reviewing other domains: a useful skill transfers even when the surface vocabulary changes completely.

Workshop 6: Museum interpretation

Source-side meaning brief: A date is proposed from stylistic similarities, not a verified document. First identify the two independent claims and the relation that the brief licenses. Keep attribution and evidence strength independent of an engaging narrative. Draft a version with as little overt marking as possible while still being clear, then draft a version with the most specific connective you can justify. Explain why each candidate does or does not preserve the original level of commitment. The point of this exercise is the decision process, not memorizing a single answer string.

Adversarial extension: imagine a target reader under time pressure who sees only the translated paragraph. Write one false inference that might be encouraged by adding because, therefore, although, only if, while or for example in the wrong place. Then replace the linker or change sentence structure so that the source meaning remains intact. Check whether the second span is an observation, reason, exception, limitation, evidence point or independent fact. If the target becomes less fluent after correcting a relation, improve the sentence while keeping the repaired relationship stable.

Completion standard: another reviewer should be able to explain why your chosen relation is supported, point to the source evidence and identify which competing rendering was excluded. Keep attribution and evidence strength independent of an engaging narrative. A source–target sheet should record the approved wording and an unresolved-author-query field. This enables later reuse without pretending that all instances of the same linking word across languages have the same discourse function. Revisit the exercise after reviewing other domains: a useful skill transfers even when the surface vocabulary changes completely.

Workshop 7: Product manual

Source-side meaning brief: One setting reduces noise, but it may decrease speed. First identify the two independent claims and the relation that the brief licenses. Preserve a trade-off and avoid transforming it into an unconditional performance claim. Draft a version with as little overt marking as possible while still being clear, then draft a version with the most specific connective you can justify. Explain why each candidate does or does not preserve the original level of commitment. The point of this exercise is the decision process, not memorizing a single answer string.

Adversarial extension: imagine a target reader under time pressure who sees only the translated paragraph. Write one false inference that might be encouraged by adding because, therefore, although, only if, while or for example in the wrong place. Then replace the linker or change sentence structure so that the source meaning remains intact. Check whether the second span is an observation, reason, exception, limitation, evidence point or independent fact. If the target becomes less fluent after correcting a relation, improve the sentence while keeping the repaired relationship stable.

Completion standard: another reviewer should be able to explain why your chosen relation is supported, point to the source evidence and identify which competing rendering was excluded. Preserve a trade-off and avoid transforming it into an unconditional performance claim. A source–target sheet should record the approved wording and an unresolved-author-query field. This enables later reuse without pretending that all instances of the same linking word across languages have the same discourse function. Revisit the exercise after reviewing other domains: a useful skill transfers even when the surface vocabulary changes completely.

Workshop 8: Literary scene

Source-side meaning brief: Someone departs, and a light remains on; the narrator offers no reason. First identify the two independent claims and the relation that the brief licenses. Preserve the opportunity for inference without naming a hidden agent. Draft a version with as little overt marking as possible while still being clear, then draft a version with the most specific connective you can justify. Explain why each candidate does or does not preserve the original level of commitment. The point of this exercise is the decision process, not memorizing a single answer string.

Adversarial extension: imagine a target reader under time pressure who sees only the translated paragraph. Write one false inference that might be encouraged by adding because, therefore, although, only if, while or for example in the wrong place. Then replace the linker or change sentence structure so that the source meaning remains intact. Check whether the second span is an observation, reason, exception, limitation, evidence point or independent fact. If the target becomes less fluent after correcting a relation, improve the sentence while keeping the repaired relationship stable.

Completion standard: another reviewer should be able to explain why your chosen relation is supported, point to the source evidence and identify which competing rendering was excluded. Preserve the opportunity for inference without naming a hidden agent. A source–target sheet should record the approved wording and an unresolved-author-query field. This enables later reuse without pretending that all instances of the same linking word across languages have the same discourse function. Revisit the exercise after reviewing other domains: a useful skill transfers even when the surface vocabulary changes completely.

Workshop 9: Emergency update

Source-side meaning brief: The power failed after a storm; investigators have not named the cause. First identify the two independent claims and the relation that the brief licenses. Keep chronological context distinct from a verified fault mechanism. Draft a version with as little overt marking as possible while still being clear, then draft a version with the most specific connective you can justify. Explain why each candidate does or does not preserve the original level of commitment. The point of this exercise is the decision process, not memorizing a single answer string.

Adversarial extension: imagine a target reader under time pressure who sees only the translated paragraph. Write one false inference that might be encouraged by adding because, therefore, although, only if, while or for example in the wrong place. Then replace the linker or change sentence structure so that the source meaning remains intact. Check whether the second span is an observation, reason, exception, limitation, evidence point or independent fact. If the target becomes less fluent after correcting a relation, improve the sentence while keeping the repaired relationship stable.

Completion standard: another reviewer should be able to explain why your chosen relation is supported, point to the source evidence and identify which competing rendering was excluded. Keep chronological context distinct from a verified fault mechanism. A source–target sheet should record the approved wording and an unresolved-author-query field. This enables later reuse without pretending that all instances of the same linking word across languages have the same discourse function. Revisit the exercise after reviewing other domains: a useful skill transfers even when the surface vocabulary changes completely.

Workshop 10: Committee minutes

Source-side meaning brief: The discussion ended, although the vote was delayed. First identify the two independent claims and the relation that the brief licenses. Preserve completed and deferred stages as separate event statuses. Draft a version with as little overt marking as possible while still being clear, then draft a version with the most specific connective you can justify. Explain why each candidate does or does not preserve the original level of commitment. The point of this exercise is the decision process, not memorizing a single answer string.

Adversarial extension: imagine a target reader under time pressure who sees only the translated paragraph. Write one false inference that might be encouraged by adding because, therefore, although, only if, while or for example in the wrong place. Then replace the linker or change sentence structure so that the source meaning remains intact. Check whether the second span is an observation, reason, exception, limitation, evidence point or independent fact. If the target becomes less fluent after correcting a relation, improve the sentence while keeping the repaired relationship stable.

Completion standard: another reviewer should be able to explain why your chosen relation is supported, point to the source evidence and identify which competing rendering was excluded. Preserve completed and deferred stages as separate event statuses. A source–target sheet should record the approved wording and an unresolved-author-query field. This enables later reuse without pretending that all instances of the same linking word across languages have the same discourse function. Revisit the exercise after reviewing other domains: a useful skill transfers even when the surface vocabulary changes completely.

Workshop 11: Press statement

Source-side meaning brief: The company says a defect was corrected; an independent test is pending. First identify the two independent claims and the relation that the brief licenses. Retain the source of the assertion and uncertainty about confirmation. Draft a version with as little overt marking as possible while still being clear, then draft a version with the most specific connective you can justify. Explain why each candidate does or does not preserve the original level of commitment. The point of this exercise is the decision process, not memorizing a single answer string.

Adversarial extension: imagine a target reader under time pressure who sees only the translated paragraph. Write one false inference that might be encouraged by adding because, therefore, although, only if, while or for example in the wrong place. Then replace the linker or change sentence structure so that the source meaning remains intact. Check whether the second span is an observation, reason, exception, limitation, evidence point or independent fact. If the target becomes less fluent after correcting a relation, improve the sentence while keeping the repaired relationship stable.

Completion standard: another reviewer should be able to explain why your chosen relation is supported, point to the source evidence and identify which competing rendering was excluded. Retain the source of the assertion and uncertainty about confirmation. A source–target sheet should record the approved wording and an unresolved-author-query field. This enables later reuse without pretending that all instances of the same linking word across languages have the same discourse function. Revisit the exercise after reviewing other domains: a useful skill transfers even when the surface vocabulary changes completely.

Workshop 12: Technical ticket

Source-side meaning brief: A test failed; the next action was to inspect a cable. First identify the two independent claims and the relation that the brief licenses. Represent diagnostic sequence without declaring the cable defective. Draft a version with as little overt marking as possible while still being clear, then draft a version with the most specific connective you can justify. Explain why each candidate does or does not preserve the original level of commitment. The point of this exercise is the decision process, not memorizing a single answer string.

Adversarial extension: imagine a target reader under time pressure who sees only the translated paragraph. Write one false inference that might be encouraged by adding because, therefore, although, only if, while or for example in the wrong place. Then replace the linker or change sentence structure so that the source meaning remains intact. Check whether the second span is an observation, reason, exception, limitation, evidence point or independent fact. If the target becomes less fluent after correcting a relation, improve the sentence while keeping the repaired relationship stable.

Completion standard: another reviewer should be able to explain why your chosen relation is supported, point to the source evidence and identify which competing rendering was excluded. Represent diagnostic sequence without declaring the cable defective. A source–target sheet should record the approved wording and an unresolved-author-query field. This enables later reuse without pretending that all instances of the same linking word across languages have the same discourse function. Revisit the exercise after reviewing other domains: a useful skill transfers even when the surface vocabulary changes completely.

Workshop 13: Learning guide

Source-side meaning brief: A complex skill has several components; the author then supplies one example. First identify the two independent claims and the relation that the brief licenses. Keep the category open and avoid converting illustration to definition. Draft a version with as little overt marking as possible while still being clear, then draft a version with the most specific connective you can justify. Explain why each candidate does or does not preserve the original level of commitment. The point of this exercise is the decision process, not memorizing a single answer string.

Adversarial extension: imagine a target reader under time pressure who sees only the translated paragraph. Write one false inference that might be encouraged by adding because, therefore, although, only if, while or for example in the wrong place. Then replace the linker or change sentence structure so that the source meaning remains intact. Check whether the second span is an observation, reason, exception, limitation, evidence point or independent fact. If the target becomes less fluent after correcting a relation, improve the sentence while keeping the repaired relationship stable.

Completion standard: another reviewer should be able to explain why your chosen relation is supported, point to the source evidence and identify which competing rendering was excluded. Keep the category open and avoid converting illustration to definition. A source–target sheet should record the approved wording and an unresolved-author-query field. This enables later reuse without pretending that all instances of the same linking word across languages have the same discourse function. Revisit the exercise after reviewing other domains: a useful skill transfers even when the surface vocabulary changes completely.

Workshop 14: Archive catalogue

Source-side meaning brief: A new record differs from an earlier label; a revision is under review. First identify the two independent claims and the relation that the brief licenses. Separate discovery, interpretation and publication status. Draft a version with as little overt marking as possible while still being clear, then draft a version with the most specific connective you can justify. Explain why each candidate does or does not preserve the original level of commitment. The point of this exercise is the decision process, not memorizing a single answer string.

Adversarial extension: imagine a target reader under time pressure who sees only the translated paragraph. Write one false inference that might be encouraged by adding because, therefore, although, only if, while or for example in the wrong place. Then replace the linker or change sentence structure so that the source meaning remains intact. Check whether the second span is an observation, reason, exception, limitation, evidence point or independent fact. If the target becomes less fluent after correcting a relation, improve the sentence while keeping the repaired relationship stable.

Completion standard: another reviewer should be able to explain why your chosen relation is supported, point to the source evidence and identify which competing rendering was excluded. Separate discovery, interpretation and publication status. A source–target sheet should record the approved wording and an unresolved-author-query field. This enables later reuse without pretending that all instances of the same linking word across languages have the same discourse function. Revisit the exercise after reviewing other domains: a useful skill transfers even when the surface vocabulary changes completely.

Workshop 15: Event invitation

Source-side meaning brief: The venue can host a session if enough attendees register. First identify the two independent claims and the relation that the brief licenses. Preserve the conditional trigger without promising the session. Draft a version with as little overt marking as possible while still being clear, then draft a version with the most specific connective you can justify. Explain why each candidate does or does not preserve the original level of commitment. The point of this exercise is the decision process, not memorizing a single answer string.

Adversarial extension: imagine a target reader under time pressure who sees only the translated paragraph. Write one false inference that might be encouraged by adding because, therefore, although, only if, while or for example in the wrong place. Then replace the linker or change sentence structure so that the source meaning remains intact. Check whether the second span is an observation, reason, exception, limitation, evidence point or independent fact. If the target becomes less fluent after correcting a relation, improve the sentence while keeping the repaired relationship stable.

Completion standard: another reviewer should be able to explain why your chosen relation is supported, point to the source evidence and identify which competing rendering was excluded. Preserve the conditional trigger without promising the session. A source–target sheet should record the approved wording and an unresolved-author-query field. This enables later reuse without pretending that all instances of the same linking word across languages have the same discourse function. Revisit the exercise after reviewing other domains: a useful skill transfers even when the surface vocabulary changes completely.

Workshop 16: Scientific caption

Source-side meaning brief: Two curves cross at one point; their convergence is not necessarily causal. First identify the two independent claims and the relation that the brief licenses. Keep visual coincidence apart from physical explanation. Draft a version with as little overt marking as possible while still being clear, then draft a version with the most specific connective you can justify. Explain why each candidate does or does not preserve the original level of commitment. The point of this exercise is the decision process, not memorizing a single answer string.

Adversarial extension: imagine a target reader under time pressure who sees only the translated paragraph. Write one false inference that might be encouraged by adding because, therefore, although, only if, while or for example in the wrong place. Then replace the linker or change sentence structure so that the source meaning remains intact. Check whether the second span is an observation, reason, exception, limitation, evidence point or independent fact. If the target becomes less fluent after correcting a relation, improve the sentence while keeping the repaired relationship stable.

Completion standard: another reviewer should be able to explain why your chosen relation is supported, point to the source evidence and identify which competing rendering was excluded. Keep visual coincidence apart from physical explanation. A source–target sheet should record the approved wording and an unresolved-author-query field. This enables later reuse without pretending that all instances of the same linking word across languages have the same discourse function. Revisit the exercise after reviewing other domains: a useful skill transfers even when the surface vocabulary changes completely.

Workshop 17: Service plan

Source-side meaning brief: Online access is included; offline storage requires a separate feature. First identify the two independent claims and the relation that the brief licenses. Ensure an exception is attached to the correct service term. Draft a version with as little overt marking as possible while still being clear, then draft a version with the most specific connective you can justify. Explain why each candidate does or does not preserve the original level of commitment. The point of this exercise is the decision process, not memorizing a single answer string.

Adversarial extension: imagine a target reader under time pressure who sees only the translated paragraph. Write one false inference that might be encouraged by adding because, therefore, although, only if, while or for example in the wrong place. Then replace the linker or change sentence structure so that the source meaning remains intact. Check whether the second span is an observation, reason, exception, limitation, evidence point or independent fact. If the target becomes less fluent after correcting a relation, improve the sentence while keeping the repaired relationship stable.

Completion standard: another reviewer should be able to explain why your chosen relation is supported, point to the source evidence and identify which competing rendering was excluded. Ensure an exception is attached to the correct service term. A source–target sheet should record the approved wording and an unresolved-author-query field. This enables later reuse without pretending that all instances of the same linking word across languages have the same discourse function. Revisit the exercise after reviewing other domains: a useful skill transfers even when the surface vocabulary changes completely.

Workshop 18: Dialog subtitle

Source-side meaning brief: The speaker says ‘yes’ to a meeting but rejects the proposed date. First identify the two independent claims and the relation that the brief licenses. Recover the local scope of agreement instead of claiming total acceptance. Draft a version with as little overt marking as possible while still being clear, then draft a version with the most specific connective you can justify. Explain why each candidate does or does not preserve the original level of commitment. The point of this exercise is the decision process, not memorizing a single answer string.

Adversarial extension: imagine a target reader under time pressure who sees only the translated paragraph. Write one false inference that might be encouraged by adding because, therefore, although, only if, while or for example in the wrong place. Then replace the linker or change sentence structure so that the source meaning remains intact. Check whether the second span is an observation, reason, exception, limitation, evidence point or independent fact. If the target becomes less fluent after correcting a relation, improve the sentence while keeping the repaired relationship stable.

Completion standard: another reviewer should be able to explain why your chosen relation is supported, point to the source evidence and identify which competing rendering was excluded. Recover the local scope of agreement instead of claiming total acceptance. A source–target sheet should record the approved wording and an unresolved-author-query field. This enables later reuse without pretending that all instances of the same linking word across languages have the same discourse function. Revisit the exercise after reviewing other domains: a useful skill transfers even when the surface vocabulary changes completely.

Workshop 19: Historical timeline

Source-side meaning brief: Two decisions happened in one year, but the source does not establish a dependency. First identify the two independent claims and the relation that the brief licenses. Preserve simultaneity and avoid causal storytelling. Draft a version with as little overt marking as possible while still being clear, then draft a version with the most specific connective you can justify. Explain why each candidate does or does not preserve the original level of commitment. The point of this exercise is the decision process, not memorizing a single answer string.

Adversarial extension: imagine a target reader under time pressure who sees only the translated paragraph. Write one false inference that might be encouraged by adding because, therefore, although, only if, while or for example in the wrong place. Then replace the linker or change sentence structure so that the source meaning remains intact. Check whether the second span is an observation, reason, exception, limitation, evidence point or independent fact. If the target becomes less fluent after correcting a relation, improve the sentence while keeping the repaired relationship stable.

Completion standard: another reviewer should be able to explain why your chosen relation is supported, point to the source evidence and identify which competing rendering was excluded. Preserve simultaneity and avoid causal storytelling. A source–target sheet should record the approved wording and an unresolved-author-query field. This enables later reuse without pretending that all instances of the same linking word across languages have the same discourse function. Revisit the exercise after reviewing other domains: a useful skill transfers even when the surface vocabulary changes completely.

Workshop 20: Consumer instructions

Source-side meaning brief: An optional accessory can improve comfort; it is not required to use the device. First identify the two independent claims and the relation that the brief licenses. Distinguish recommendation, optionality, prerequisite and consequence. Draft a version with as little overt marking as possible while still being clear, then draft a version with the most specific connective you can justify. Explain why each candidate does or does not preserve the original level of commitment. The point of this exercise is the decision process, not memorizing a single answer string.

Adversarial extension: imagine a target reader under time pressure who sees only the translated paragraph. Write one false inference that might be encouraged by adding because, therefore, although, only if, while or for example in the wrong place. Then replace the linker or change sentence structure so that the source meaning remains intact. Check whether the second span is an observation, reason, exception, limitation, evidence point or independent fact. If the target becomes less fluent after correcting a relation, improve the sentence while keeping the repaired relationship stable.

Completion standard: another reviewer should be able to explain why your chosen relation is supported, point to the source evidence and identify which competing rendering was excluded. Distinguish recommendation, optionality, prerequisite and consequence. A source–target sheet should record the approved wording and an unresolved-author-query field. This enables later reuse without pretending that all instances of the same linking word across languages have the same discourse function. Revisit the exercise after reviewing other domains: a useful skill transfers even when the surface vocabulary changes completely.

Workshop 21: Research discussion

Source-side meaning brief: An effect is statistically detectable but the mechanism remains unknown. First identify the two independent claims and the relation that the brief licenses. Preserve the strength of each claim and the remaining unknown. Draft a version with as little overt marking as possible while still being clear, then draft a version with the most specific connective you can justify. Explain why each candidate does or does not preserve the original level of commitment. The point of this exercise is the decision process, not memorizing a single answer string.

Adversarial extension: imagine a target reader under time pressure who sees only the translated paragraph. Write one false inference that might be encouraged by adding because, therefore, although, only if, while or for example in the wrong place. Then replace the linker or change sentence structure so that the source meaning remains intact. Check whether the second span is an observation, reason, exception, limitation, evidence point or independent fact. If the target becomes less fluent after correcting a relation, improve the sentence while keeping the repaired relationship stable.

Completion standard: another reviewer should be able to explain why your chosen relation is supported, point to the source evidence and identify which competing rendering was excluded. Preserve the strength of each claim and the remaining unknown. A source–target sheet should record the approved wording and an unresolved-author-query field. This enables later reuse without pretending that all instances of the same linking word across languages have the same discourse function. Revisit the exercise after reviewing other domains: a useful skill transfers even when the surface vocabulary changes completely.

Workshop 22: Financial explanation

Source-side meaning brief: A discount is available if criteria are met; no approval has yet been issued. First identify the two independent claims and the relation that the brief licenses. Separate eligibility rule from administrative confirmation. Draft a version with as little overt marking as possible while still being clear, then draft a version with the most specific connective you can justify. Explain why each candidate does or does not preserve the original level of commitment. The point of this exercise is the decision process, not memorizing a single answer string.

Adversarial extension: imagine a target reader under time pressure who sees only the translated paragraph. Write one false inference that might be encouraged by adding because, therefore, although, only if, while or for example in the wrong place. Then replace the linker or change sentence structure so that the source meaning remains intact. Check whether the second span is an observation, reason, exception, limitation, evidence point or independent fact. If the target becomes less fluent after correcting a relation, improve the sentence while keeping the repaired relationship stable.

Completion standard: another reviewer should be able to explain why your chosen relation is supported, point to the source evidence and identify which competing rendering was excluded. Separate eligibility rule from administrative confirmation. A source–target sheet should record the approved wording and an unresolved-author-query field. This enables later reuse without pretending that all instances of the same linking word across languages have the same discourse function. Revisit the exercise after reviewing other domains: a useful skill transfers even when the surface vocabulary changes completely.

Workshop 23: Environment report

Source-side meaning brief: Water quality improved and a restoration project was underway. First identify the two independent claims and the relation that the brief licenses. Avoid claiming intervention effectiveness without supporting comparison. Draft a version with as little overt marking as possible while still being clear, then draft a version with the most specific connective you can justify. Explain why each candidate does or does not preserve the original level of commitment. The point of this exercise is the decision process, not memorizing a single answer string.

Adversarial extension: imagine a target reader under time pressure who sees only the translated paragraph. Write one false inference that might be encouraged by adding because, therefore, although, only if, while or for example in the wrong place. Then replace the linker or change sentence structure so that the source meaning remains intact. Check whether the second span is an observation, reason, exception, limitation, evidence point or independent fact. If the target becomes less fluent after correcting a relation, improve the sentence while keeping the repaired relationship stable.

Completion standard: another reviewer should be able to explain why your chosen relation is supported, point to the source evidence and identify which competing rendering was excluded. Avoid claiming intervention effectiveness without supporting comparison. A source–target sheet should record the approved wording and an unresolved-author-query field. This enables later reuse without pretending that all instances of the same linking word across languages have the same discourse function. Revisit the exercise after reviewing other domains: a useful skill transfers even when the surface vocabulary changes completely.

Workshop 24: Medical research summary

Source-side meaning brief: Participants reported improvement; the study lacked a control group. First identify the two independent claims and the relation that the brief licenses. Keep the limitations visible without declaring the intervention either effective or ineffective. Draft a version with as little overt marking as possible while still being clear, then draft a version with the most specific connective you can justify. Explain why each candidate does or does not preserve the original level of commitment. The point of this exercise is the decision process, not memorizing a single answer string.

Adversarial extension: imagine a target reader under time pressure who sees only the translated paragraph. Write one false inference that might be encouraged by adding because, therefore, although, only if, while or for example in the wrong place. Then replace the linker or change sentence structure so that the source meaning remains intact. Check whether the second span is an observation, reason, exception, limitation, evidence point or independent fact. If the target becomes less fluent after correcting a relation, improve the sentence while keeping the repaired relationship stable.

Completion standard: another reviewer should be able to explain why your chosen relation is supported, point to the source evidence and identify which competing rendering was excluded. Keep the limitations visible without declaring the intervention either effective or ineffective. A source–target sheet should record the approved wording and an unresolved-author-query field. This enables later reuse without pretending that all instances of the same linking word across languages have the same discourse function. Revisit the exercise after reviewing other domains: a useful skill transfers even when the surface vocabulary changes completely.

7. A complete quality-control worksheet

• Document ID and version: capture the precise source text, intended audience and applicable specifications.
• Source span A: preserve its proposition, speaker, certainty and date or condition.
• Source span B: preserve its proposition and its relationship to span A.
• Source-side marker: quote the actual connective or lexical signal, or record ‘implicit’.
• Candidate relation labels: write primary and alternative interpretations, if needed.
• Scope and direction: record whether the link runs A to B, B to A, or across a larger paragraph.
• Evidence: distinguish encoded grammar, contextual implication, genre convention and author clarification.
• Minimum target: draft a clear target version with no stronger commitment than the source.
• Explicit target: draft a version with a relation marker when source evidence allows it.
• Changed inference: identify any new claim, condition, expectation or explanation.
• Reader action: identify what an intended reader would do differently.
• Author query: specify one factual question that resolves a material ambiguity.
• Reviewer status: classify as approved, needs change, needs author confirmation or intentionally ambiguous.
• Change history: log when a source or translation revision changes the relation map.
• Release test: ask an independent reader to paraphrase both propositions and the relationship they infer.

8. A four-week training and implementation path

Week one: notice the bridge

Collect short authentic passages with two adjacent ideas. Do not translate them immediately. Annotate cause, contrast, concession, time, explanation, example and uncertainty separately; justify each label with source evidence. If a passage remains ambiguous, annotate its competing possibilities. The objective is accurate recognition, not speed.

At the end of this stage, collect evidence of improved decisions: annotated spans, justified connective choices, corrected false inferences, and independent reader answers. Avoid measuring success merely by the number of explicit discourse markers inserted or removed. A good revision protects the intended relationship, preserves uncertainty where appropriate, and produces a target document that works in the reader’s real situation.

Week two: write two defensible targets

For every short passage, create a minimal target version and a more explicit version. Compare what each version commits the author to. Reject extra causal, conditional or evidential assertions. Read aloud to check that faithful wording remains natural and understandable in the target language.

At the end of this stage, collect evidence of improved decisions: annotated spans, justified connective choices, corrected false inferences, and independent reader answers. Avoid measuring success merely by the number of explicit discourse markers inserted or removed. A good revision protects the intended relationship, preserves uncertainty where appropriate, and produces a target document that works in the reader’s real situation.

Week three: run adversarial reader tests

Have a peer read target-only text and answer questions such as ‘Was the first event proved to cause the second?’, ‘Was this an example or the full list?’, ‘Was the exception binding?’, and ‘Who asserted the explanation?’. Compare answers to the source-side relation map, and investigate mismatches rather than counting linkers.

At the end of this stage, collect evidence of improved decisions: annotated spans, justified connective choices, corrected false inferences, and independent reader answers. Avoid measuring success merely by the number of explicit discourse markers inserted or removed. A good revision protects the intended relationship, preserves uncertainty where appropriate, and produces a target document that works in the reader’s real situation.

Week four: implement a release gate

Add relation mapping to translation briefs, translation-memory context notes, human review and post-editing. Prioritize high-risk genres and add a log for author clarifications. Review a small sample across complete paragraphs after changes. Record recurring error types so future translators learn from actual failures rather than generic style warnings.

At the end of this stage, collect evidence of improved decisions: annotated spans, justified connective choices, corrected false inferences, and independent reader answers. Avoid measuring success merely by the number of explicit discourse markers inserted or removed. A good revision protects the intended relationship, preserves uncertainty where appropriate, and produces a target document that works in the reader’s real situation.

9. Frequently asked questions

Should translators always add connectives to make implicit relations clear?

No. Adding a marker can help when the relationship is established and a target reader needs guidance. It can also manufacture causality, concession or certainty. The decision depends on source commitment, genre and reader needs, not a blanket preference for explicit prose.

Practical check: isolate the proposition pair, write down the inferred relationship, and compare a stronger and a weaker connective. Explain what extra commitment the stronger form adds. The purpose is to make the translation decision inspectable, not to reward a mechanically literal surface. Where the source does not resolve a consequential difference, preserve it or obtain a verified answer rather than guessing.

Is it wrong to translate two sentences as one sentence?

Not automatically. Sentence boundaries can change while meaning is preserved. The reviewer must check what the chosen coordination or subordination asserts, especially whether a subordinate because, although, unless or if clause adds a new relation.

Practical check: isolate the proposition pair, write down the inferred relationship, and compare a stronger and a weaker connective. Explain what extra commitment the stronger form adds. The purpose is to make the translation decision inspectable, not to reward a mechanically literal surface. Where the source does not resolve a consequential difference, preserve it or obtain a verified answer rather than guessing.

What is the difference between contrast and concession?

Contrast presents comparable differences; concession creates an expectation that one fact would make another less likely or surprising. Some sentences support both readings, but translators should not impose the more forceful one without evidence.

Practical check: isolate the proposition pair, write down the inferred relationship, and compare a stronger and a weaker connective. Explain what extra commitment the stronger form adds. The purpose is to make the translation decision inspectable, not to reward a mechanically literal surface. Where the source does not resolve a consequential difference, preserve it or obtain a verified answer rather than guessing.

Why is therefore more dangerous than and?

Therefore explicitly invites a consequence or inference from what preceded it. And is often less committed, although its pragmatic implications still depend on context. The stronger relation may be precisely what the author intends, but it must be justified.

Practical check: isolate the proposition pair, write down the inferred relationship, and compare a stronger and a weaker connective. Explain what extra commitment the stronger form adds. The purpose is to make the translation decision inspectable, not to reward a mechanically literal surface. Where the source does not resolve a consequential difference, preserve it or obtain a verified answer rather than guessing.

Can a translator preserve ambiguity?

Yes. Natural syntax, punctuation or neutral coordination can preserve a source passage’s interpretive openness. Material ambiguity affecting action, rights or safety may instead need an author query before release.

Practical check: isolate the proposition pair, write down the inferred relationship, and compare a stronger and a weaker connective. Explain what extra commitment the stronger form adds. The purpose is to make the translation decision inspectable, not to reward a mechanically literal surface. Where the source does not resolve a consequential difference, preserve it or obtain a verified answer rather than guessing.

What if the source connective itself is ambiguous?

Use syntax, semantic content, paragraph purpose and speaker context to identify the most likely function. If multiple interpretations remain plausible and materially differ, record them and seek qualified clarification.

Practical check: isolate the proposition pair, write down the inferred relationship, and compare a stronger and a weaker connective. Explain what extra commitment the stronger form adds. The purpose is to make the translation decision inspectable, not to reward a mechanically literal surface. Where the source does not resolve a consequential difference, preserve it or obtain a verified answer rather than guessing.

Is coherence the same as cohesion?

Not exactly. Coherence is an intelligible relationship between ideas; cohesion includes visible textual resources such as pronouns and connectives. Cohesive writing can still be incoherent, and coherent writing can use implicit links.

Practical check: isolate the proposition pair, write down the inferred relationship, and compare a stronger and a weaker connective. Explain what extra commitment the stronger form adds. The purpose is to make the translation decision inspectable, not to reward a mechanically literal surface. Where the source does not resolve a consequential difference, preserve it or obtain a verified answer rather than guessing.

How does this matter for machine translation?

Fluent automatic output may insert or remove relation signals. Human reviewers should compare the source and target relation maps, focusing on unsupported cause, purpose, contrast, conditional triggers and evidential claims.

Practical check: isolate the proposition pair, write down the inferred relationship, and compare a stronger and a weaker connective. Explain what extra commitment the stronger form adds. The purpose is to make the translation decision inspectable, not to reward a mechanically literal surface. Where the source does not resolve a consequential difference, preserve it or obtain a verified answer rather than guessing.

Can back-translation catch errors?

It can reveal some meaning shifts but is not decisive. Two translators can reproduce the same plausible wrong inference. Direct source–target checking and independent reader comprehension tests are stronger when used together.

Practical check: isolate the proposition pair, write down the inferred relationship, and compare a stronger and a weaker connective. Explain what extra commitment the stronger form adds. The purpose is to make the translation decision inspectable, not to reward a mechanically literal surface. Where the source does not resolve a consequential difference, preserve it or obtain a verified answer rather than guessing.

Can translation memory reuse a connective decision?

Only with appropriate context. A segment translated in a causal paragraph may occur later in an additive or temporal paragraph. Translation memories should preserve enough surrounding information and revision history for reviewers to re-evaluate the link.

Practical check: isolate the proposition pair, write down the inferred relationship, and compare a stronger and a weaker connective. Explain what extra commitment the stronger form adds. The purpose is to make the translation decision inspectable, not to reward a mechanically literal surface. Where the source does not resolve a consequential difference, preserve it or obtain a verified answer rather than guessing.

Do language communities always prefer different numbers of connectives?

Usage patterns vary by language pair, genre, register and individual style. Do not impose national stereotypes. Consult suitable parallel texts and the actual translation brief before generalizing.

Practical check: isolate the proposition pair, write down the inferred relationship, and compare a stronger and a weaker connective. Explain what extra commitment the stronger form adds. The purpose is to make the translation decision inspectable, not to reward a mechanically literal surface. Where the source does not resolve a consequential difference, preserve it or obtain a verified answer rather than guessing.

What if the translated version sounds unnatural without an explicit marker?

Try clause reordering, punctuation, a topic phrase or a weaker marker that does not overstate meaning. Fluency and fidelity need not be opposing goals. When the target requires contextual help, distinguish authorized explanation from source assertion.

Practical check: isolate the proposition pair, write down the inferred relationship, and compare a stronger and a weaker connective. Explain what extra commitment the stronger form adds. The purpose is to make the translation decision inspectable, not to reward a mechanically literal surface. Where the source does not resolve a consequential difference, preserve it or obtain a verified answer rather than guessing.

Are all if-statements conditions?

They can vary in rhetorical or pragmatic function, but their conditional commitments must be analyzed. A condition about possible processing is not the same as a forecast that processing will occur or a statement that it already occurred.

Practical check: isolate the proposition pair, write down the inferred relationship, and compare a stronger and a weaker connective. Explain what extra commitment the stronger form adds. The purpose is to make the translation decision inspectable, not to reward a mechanically literal surface. Where the source does not resolve a consequential difference, preserve it or obtain a verified answer rather than guessing.

How should a reviewer report a relation problem?

Quote the source and target spans, identify the changed inference, record its severity and propose a minimal repair. Where interpretation remains open, ask a focused author question rather than making a hidden unilateral decision.

Practical check: isolate the proposition pair, write down the inferred relationship, and compare a stronger and a weaker connective. Explain what extra commitment the stronger form adds. The purpose is to make the translation decision inspectable, not to reward a mechanically literal surface. Where the source does not resolve a consequential difference, preserve it or obtain a verified answer rather than guessing.

How do examples and lists get mistranslated?

Illustrative lists can become exhaustive when rendered with only or namely; exhaustive sets can become open when rendered as examples. Check whether the source describes instances, required steps, options or the complete membership of a category.

Practical check: isolate the proposition pair, write down the inferred relationship, and compare a stronger and a weaker connective. Explain what extra commitment the stronger form adds. The purpose is to make the translation decision inspectable, not to reward a mechanically literal surface. Where the source does not resolve a consequential difference, preserve it or obtain a verified answer rather than guessing.

What should be tested in subtitles?

Check limited context, speaker changes, timing, and whether a short connector gives the correct contrast or concession. Tight space is not a license to collapse a conditional or assign an unstated cause.

Practical check: isolate the proposition pair, write down the inferred relationship, and compare a stronger and a weaker connective. Explain what extra commitment the stronger form adds. The purpose is to make the translation decision inspectable, not to reward a mechanically literal surface. Where the source does not resolve a consequential difference, preserve it or obtain a verified answer rather than guessing.

Does a discourse relation error count as a translation quality error?

It can, when it violates project requirements or affects intended meaning. ISO 5060:2024 provides general guidance for human evaluation using error types and severity; it does not dictate one universal connective taxonomy.

Practical check: isolate the proposition pair, write down the inferred relationship, and compare a stronger and a weaker connective. Explain what extra commitment the stronger form adds. The purpose is to make the translation decision inspectable, not to reward a mechanically literal surface. Where the source does not resolve a consequential difference, preserve it or obtain a verified answer rather than guessing.

How does a reader comprehension test work?

Give target-only readers practical questions about what caused an event, what was merely observed, what condition applies and which options are included. Compare their answers with source-authorized meaning, then revise misleading links.

Practical check: isolate the proposition pair, write down the inferred relationship, and compare a stronger and a weaker connective. Explain what extra commitment the stronger form adds. The purpose is to make the translation decision inspectable, not to reward a mechanically literal surface. Where the source does not resolve a consequential difference, preserve it or obtain a verified answer rather than guessing.

How do you translate a research limitation without overstating it?

Keep observed results distinct from inferential limitations. A small sample may justify cautious interpretation, not the conclusion that a study is worthless; lack of proof of causation does not itself prove an effect absent.

Practical check: isolate the proposition pair, write down the inferred relationship, and compare a stronger and a weaker connective. Explain what extra commitment the stronger form adds. The purpose is to make the translation decision inspectable, not to reward a mechanically literal surface. Where the source does not resolve a consequential difference, preserve it or obtain a verified answer rather than guessing.

What is the fastest daily exercise?

Take one two-sentence passage and write three interpretations: temporal sequence, causal explanation and concession. Identify which are supported and which require added premises. Then produce one natural, evidence-calibrated target translation.

Practical check: isolate the proposition pair, write down the inferred relationship, and compare a stronger and a weaker connective. Explain what extra commitment the stronger form adds. The purpose is to make the translation decision inspectable, not to reward a mechanically literal surface. Where the source does not resolve a consequential difference, preserve it or obtain a verified answer rather than guessing.

10. Terminology and concept distinctions

Discourse relation

A relationship between two or more spans that helps a reader interpret their combined role in a text.

Use discourse relation as a diagnostic category, not a magic substitute for context. The translator still needs to identify the actual source span, the target wording and the reader-relevant interpretation. Where the technical term belongs to a particular theory, note the framework instead of pretending every taxonomy is identical.

Implicit relation

A link inferred without an overt conventional connective at that point in the source.

Use implicit relation as a diagnostic category, not a magic substitute for context. The translator still needs to identify the actual source span, the target wording and the reader-relevant interpretation. Where the technical term belongs to a particular theory, note the framework instead of pretending every taxonomy is identical.

Explicit connective

A lexical or grammatical signal such as because, however, although or for example.

Use explicit connective as a diagnostic category, not a magic substitute for context. The translator still needs to identify the actual source span, the target wording and the reader-relevant interpretation. Where the technical term belongs to a particular theory, note the framework instead of pretending every taxonomy is identical.

Alternative lexicalization

A noncanonical phrase that expresses a discourse link, such as ‘this explains the delay’.

Use alternative lexicalization as a diagnostic category, not a magic substitute for context. The translator still needs to identify the actual source span, the target wording and the reader-relevant interpretation. Where the technical term belongs to a particular theory, note the framework instead of pretending every taxonomy is identical.

Contingency

A family covering causes, results, reasons and conditions in some discourse frameworks.

Use contingency as a diagnostic category, not a magic substitute for context. The translator still needs to identify the actual source span, the target wording and the reader-relevant interpretation. Where the technical term belongs to a particular theory, note the framework instead of pretending every taxonomy is identical.

Comparison

A family of similarities, contrasts and expectation-defeating differences.

Use comparison as a diagnostic category, not a magic substitute for context. The translator still needs to identify the actual source span, the target wording and the reader-relevant interpretation. Where the technical term belongs to a particular theory, note the framework instead of pretending every taxonomy is identical.

Expansion

A family in which a span adds, specifies, illustrates or restates other content.

Use expansion as a diagnostic category, not a magic substitute for context. The translator still needs to identify the actual source span, the target wording and the reader-relevant interpretation. Where the technical term belongs to a particular theory, note the framework instead of pretending every taxonomy is identical.

Temporal relation

A relationship involving sequence, simultaneity or other event ordering.

Use temporal relation as a diagnostic category, not a magic substitute for context. The translator still needs to identify the actual source span, the target wording and the reader-relevant interpretation. Where the technical term belongs to a particular theory, note the framework instead of pretending every taxonomy is identical.

Concession

A relation in which one fact holds despite an expectation established by another.

Use concession as a diagnostic category, not a magic substitute for context. The translator still needs to identify the actual source span, the target wording and the reader-relevant interpretation. Where the technical term belongs to a particular theory, note the framework instead of pretending every taxonomy is identical.

Contrast

A comparison between differing aspects or positions, not necessarily a logical contradiction.

Use contrast as a diagnostic category, not a magic substitute for context. The translator still needs to identify the actual source span, the target wording and the reader-relevant interpretation. Where the technical term belongs to a particular theory, note the framework instead of pretending every taxonomy is identical.

Purpose

A relation connecting an action to an intended goal, not necessarily an attained result.

Use purpose as a diagnostic category, not a magic substitute for context. The translator still needs to identify the actual source span, the target wording and the reader-relevant interpretation. Where the technical term belongs to a particular theory, note the framework instead of pretending every taxonomy is identical.

Justification

A link providing reasons to accept an action, claim or decision as appropriate.

Use justification as a diagnostic category, not a magic substitute for context. The translator still needs to identify the actual source span, the target wording and the reader-relevant interpretation. Where the technical term belongs to a particular theory, note the framework instead of pretending every taxonomy is identical.

Explanation

A link giving reasons why an event or state occurred or is described as it is.

Use explanation as a diagnostic category, not a magic substitute for context. The translator still needs to identify the actual source span, the target wording and the reader-relevant interpretation. Where the technical term belongs to a particular theory, note the framework instead of pretending every taxonomy is identical.

Elaboration

A span supplies further detail about another span without necessarily serving as proof.

Use elaboration as a diagnostic category, not a magic substitute for context. The translator still needs to identify the actual source span, the target wording and the reader-relevant interpretation. Where the technical term belongs to a particular theory, note the framework instead of pretending every taxonomy is identical.

Instantiation

A span supplies an instance of a generalization or category.

Use instantiation as a diagnostic category, not a magic substitute for context. The translator still needs to identify the actual source span, the target wording and the reader-relevant interpretation. Where the technical term belongs to a particular theory, note the framework instead of pretending every taxonomy is identical.

Condition

A proposition that restricts when an outcome, rule or obligation applies.

Use condition as a diagnostic category, not a magic substitute for context. The translator still needs to identify the actual source span, the target wording and the reader-relevant interpretation. Where the technical term belongs to a particular theory, note the framework instead of pretending every taxonomy is identical.

Scope

The portion of text or proposition governed by a marker or relation.

Use scope as a diagnostic category, not a magic substitute for context. The translator still needs to identify the actual source span, the target wording and the reader-relevant interpretation. Where the technical term belongs to a particular theory, note the framework instead of pretending every taxonomy is identical.

Nuclearity

Within Rhetorical Structure Theory, a way of distinguishing the more central span from its supporting span.

Use nuclearity as a diagnostic category, not a magic substitute for context. The translator still needs to identify the actual source span, the target wording and the reader-relevant interpretation. Where the technical term belongs to a particular theory, note the framework instead of pretending every taxonomy is identical.

Explicitation

Making information or relations more overt in a translation than in the source expression.

Use explicitation as a diagnostic category, not a magic substitute for context. The translator still needs to identify the actual source span, the target wording and the reader-relevant interpretation. Where the technical term belongs to a particular theory, note the framework instead of pretending every taxonomy is identical.

Implicitation

Rendering a relation less overt while potentially keeping it recoverable in context.

Use implicitation as a diagnostic category, not a magic substitute for context. The translator still needs to identify the actual source span, the target wording and the reader-relevant interpretation. Where the technical term belongs to a particular theory, note the framework instead of pretending every taxonomy is identical.

Relation reversal

A mistake that changes the direction of a cause, condition, explanation or supporting link.

Use relation reversal as a diagnostic category, not a magic substitute for context. The translator still needs to identify the actual source span, the target wording and the reader-relevant interpretation. Where the technical term belongs to a particular theory, note the framework instead of pretending every taxonomy is identical.

Evidential commitment

How strongly the discourse presents a claim as supported, asserted or attributed.

Use evidential commitment as a diagnostic category, not a magic substitute for context. The translator still needs to identify the actual source span, the target wording and the reader-relevant interpretation. Where the technical term belongs to a particular theory, note the framework instead of pretending every taxonomy is identical.

Source-side brief

A record of intended meaning and context that constrains translation choices.

Use source-side brief as a diagnostic category, not a magic substitute for context. The translator still needs to identify the actual source span, the target wording and the reader-relevant interpretation. Where the technical term belongs to a particular theory, note the framework instead of pretending every taxonomy is identical.

Reader action test

A check of what a target reader would do or conclude after reading a translation.

Use reader action test as a diagnostic category, not a magic substitute for context. The translator still needs to identify the actual source span, the target wording and the reader-relevant interpretation. Where the technical term belongs to a particular theory, note the framework instead of pretending every taxonomy is identical.

Author query

A narrowly defined request to resolve missing or conflicting source information.

Use author query as a diagnostic category, not a magic substitute for context. The translator still needs to identify the actual source span, the target wording and the reader-relevant interpretation. Where the technical term belongs to a particular theory, note the framework instead of pretending every taxonomy is identical.

Semantic invariant

A content or relation property the translation must preserve despite changes in wording.

Use semantic invariant as a diagnostic category, not a magic substitute for context. The translator still needs to identify the actual source span, the target wording and the reader-relevant interpretation. Where the technical term belongs to a particular theory, note the framework instead of pretending every taxonomy is identical.

Inference upgrade

An unsupported strengthening from possibility, association or sequence to a more assertive claim.

Use inference upgrade as a diagnostic category, not a magic substitute for context. The translator still needs to identify the actual source span, the target wording and the reader-relevant interpretation. Where the technical term belongs to a particular theory, note the framework instead of pretending every taxonomy is identical.

Translationese

Target prose strongly shaped by source-language patterns and sometimes awkward or misleading.

Use translationese as a diagnostic category, not a magic substitute for context. The translator still needs to identify the actual source span, the target wording and the reader-relevant interpretation. Where the technical term belongs to a particular theory, note the framework instead of pretending every taxonomy is identical.

Relation map

A structured representation of discourse spans, their direction and the links among them.

Use relation map as a diagnostic category, not a magic substitute for context. The translator still needs to identify the actual source span, the target wording and the reader-relevant interpretation. Where the technical term belongs to a particular theory, note the framework instead of pretending every taxonomy is identical.

Adversarial paraphrase

A deliberate attempt to expose hidden changes by re-expressing what the target logically commits to.

Use adversarial paraphrase as a diagnostic category, not a magic substitute for context. The translator still needs to identify the actual source span, the target wording and the reader-relevant interpretation. Where the technical term belongs to a particular theory, note the framework instead of pretending every taxonomy is identical.

11. Research, standards and further reading

The research and standards below are selected for their relevance to discourse relations, explicitation and quality evaluation. They support the analytical approach, not a claim that one standardized workflow or numerical score guarantees perfect translation. Research about one language pair should not be generalized without qualification to all languages.

Yung et al. (2023), Investigating Explicitation of Discourse Connectives in Translation using Automatic Annotations, ACL Anthology

Cambridge University Press, Connectives and Discourse Relations — Theoretical Models of Discourse Relations

ISO 5060:2024, Translation services — Evaluation of translation output — General guidance

W3C, Internationalization Tag Set Version 2.0 (29 October 2013)

Pollkläsener et al. (2024), Capturing variation of discourse relations in English parallel data through automatic annotation and alignment

Zufferey (2016), Discourse connectives across languages

12. Next learning routes and the place of this chapter

The implicit coherence problem sits between sentence-level semantics and whole-document reasoning. The previous chapter covers time anchors: a translator must know when events happen before inferring what caused what. A different chapter on arguments can explain evidence and warrants; this chapter focuses specifically on whether a source makes the connection overt, leaves it inferable, or deliberately refuses to commit. These are complementary tasks, not competing articles.

Return to the translation root: Master Art of Translation | The Complete System for Moving Meaning Between Languages

Previous reserved-tens chapter (1500): Temporal Anchor Fidelity

Protected Vocabulary Learning Hub — words, concepts and linking-language foundations

Protected How English Works — Translation and Meaning Across Languages

The practical takeaway is neither ‘add more linking words’ nor ‘never add a linking word.’ Translate what the two clauses say, what relationship the source licenses, how explicitly it licenses that relationship, and what target readers are entitled to infer. Repeat this analysis whenever fluent target prose introduces an apparently harmless because, however, while, therefore or although. Protecting an invisible bridge is often what turns a grammatically correct translation into an honest one.

Editorial provenance: EDKSG-TRANS-MASTER-WORLD-1510. This is an independently constructed world-facing instructional chapter. It does not revise its root or neighboring chapters and does not claim that its examples are certified translations from specific foreign-language originals. The techniques are educational aids; critical legal, clinical, regulatory or safety translations require appropriate professional review and source-authoritative verification.