Master Art of Translation | The Participant Role Integrity System — How Agents, Patients, Experiencers, Instruments and Voice Preserve Who Did What

How do you translate passive voice, active voice, agent and patient roles, or sentences in which one person causes something to happen to another? Accurate translation must preserve who acts, who is affected, who experiences an event, what instrument is used, whether an action is deliberate, and whether the source actually identifies the responsible party. This practical guide explains how to translate “who did what to whom” between languages without turning a patient into an agent, inventing an unnamed actor, or changing a cautious record into an accusation.

Readers searching for semantic roles in translation, agent versus patient in grammar, how to translate passive voice accurately, causative constructions across languages, thematic roles and translation quality, or active-to-passive translation examples usually need more than a grammar rule. The hidden problem is that grammatical subjects do not always perform actions: “The valve was closed,” “The valve closed,” and “The technician closed the valve” describe different commitments, even where the same object ends in the same state. Fluent wording can quietly add a technician, erase a decision, imply deliberate action, or attribute blame.

The Participant Role Integrity System gives translators, interpreters, localization specialists, editors and quality reviewers a repeatable procedure: construct an event-and-participant ledger, separate observable roles from inferred roles, map each language’s voice and argument structure, choose a natural target-language expression, then test the translation with counterexamples. The method belongs to the Master Art of Translation root. It specifically owns cross-language participant-role fidelity; it does not replace general English voice instruction, vocabulary study, legal judgment, technical expertise, or earlier chapters on causal inference, reference chains and discourse focus.

Architecture reference: EDKSG-TRANS-MASTER-WORLD-1430. Global audience; no location or jurisdiction assumption. Examples are deliberately invented English-language meaning glosses, not claims that a particular target language must use one grammatical construction. Before release, a qualified reviewer must check the actual source and target languages and the intended use. Last editorial research review: 4 November 2026.

Quick answer: preserve participants before preserving word order

For every material event, write down the event, actor or cause, affected participant, experiencer, recipient, instrument, result, agency or volition, source of knowledge, and any roles deliberately left unknown. Compare the same fields after translating. A translation passes only when its ordinary reader would identify the same relationships and would not infer an additional actor, intention or obligation that the source does not support.

Agent: typically an intentional instigator, but not every grammatical subject is an agent.
Patient: a participant that undergoes a change or is affected; it can be the subject of a passive sentence.
Theme: an entity located, moved, possessed or otherwise involved without necessarily being changed.
Experiencer: someone who perceives, feels, remembers or understands, not necessarily a voluntary actor.
Instrument: the means used in an event; do not treat it automatically as the person responsible.
Causer: an entity or condition initiating an event; causation is not the same as intent.
Recipient or beneficiary: someone who receives an object, information or benefit; a recipient is not always an agent.
Unknown role: unknown must remain unknown, rather than be supplied by translator imagination.

Contents and reading routes

The hidden problem: voice, event roles and the illusion of a faithful sentence
The mechanism: event frames, argument structure and cross-language meaning
The role ledger and eighteen-pass translation method
Role tests: truth conditions, omission, agency and recoverability
Failure modes and severity-based review
Worked diagnostics across different communication domains
Transfer workshops and practical editing drills
Specialist glossary
Frequently asked questions
Research and terminology references
Next learning routes

The hidden problem: when grammatical fluency changes the event

Imagine the message “The request was rejected.” That sentence confirms an outcome and syntactically leaves the rejecting agent unexpressed. An editor might prefer “The manager rejected the request,” thinking that a more direct sentence will help readers. But that change adds a manager. Without evidence, the new version is not more accurate: it commits the source to a person or role that it did not identify. Conversely, if the original explicitly says “The manager rejected the request,” translating it as “The request was rejected” can suppress accountability where the actor is material. The correct choice is driven by evidence and communicative function, not by a blanket preference for active or passive voice.

Now compare “The door opened,” “The door was opened,” “A gust opened the door,” and “Someone opened the door.” The first can report an event without choosing a cause; the second commonly suggests a caused event, potentially without identifying the agent; the third gives a nonhuman cause; the fourth asserts an unidentified someone. These sentences can be appropriate in different contexts but are not interchangeable. A language that packages change-of-state events differently may not reproduce the original English grammar. It must nevertheless preserve the available causal and participant information to the extent natural and explicit, or flag any unavoidable ambiguity.

Another hidden issue arises when translators equate a verb’s position with a person’s intention. “Mara received the file” does not necessarily say Mara asked for it. “Mara noticed the stain” does not say she deliberately inspected the surface. “Mara failed to stop the machine” does not, by itself, establish she was assigned the duty, knew the risk or had physical control. A narrative may later provide these facts, but they must not leak backward into the translation of an earlier statement. The ledger deliberately separates who participates from what the source asserts about their mental state, capability or responsibility.

Word-for-word methods also struggle with the same event across languages. Some languages permit frequent null subjects; others require an overt subject or placeholder. Some mark voice through verbal morphology, others through auxiliaries, word order, case or constructions that do not map neatly onto English. Some foreground the affected entity; others more naturally profile the process, outcome or experiencer. A grammatical substitution may be required for idiomatic writing, yet the result must keep the source’s commitments. The central question is not “Did the translator preserve passive morphology?” It is “Would a reader in the target language understand the same event and the same uncertainty about its participants?”

These distinctions matter in everyday translation, not just academic semantics. Incident reports can affect investigations. Product manuals direct a user rather than a machine. Research summaries distinguish a device’s measured effect from an operator’s deliberate intervention. Historical accounts may describe an action with unknown attribution. Customer support messages can apologize for a result without falsely alleging who caused it. Subtitles condense dialogue, where dropping a short agent phrase can change a character’s actions. Fiction depends on whether the narrator knows what another person intended. In each setting, an unmarked change to agency can alter the reader’s decision.

The mechanism: map an event, not its surface arrangement

A practical translation frame can be written in plain language: Event = CHANGE; Cause = heat; Patient = adhesive; Result = adhesive softened; Agent = not specified; Intent = not specified; Evidence = observation. The frame is not a universal grammatical theory. It is a compact annotation the editor can use to compare source and target texts. If a target version reads “The operator softened the adhesive with heat,” the reviewer can immediately see the added operator. If it reads “The adhesive dissolved,” the event’s result has changed from softening to dissolving. A fluency checker focused on spelling and syntax might miss both errors.

Semantic roles are relational rather than permanent properties of people or objects. The same person can be an agent in “Lea lifted the box,” a recipient in “Lea received the box,” an experiencer in “Lea heard a noise,” and a patient in “Lea was injured.” A termbase storing “Lea = agent” would therefore be conceptually wrong. Roles belong to the event frame. The frame may have two actions and require the person to have different roles in each: “Lea asked Ravi to carry the box.” Lea is the asker; Ravi is the prospective carrier; the box is what Ravi is supposed to carry. Simplifying to “Lea carried the box” is a severe transfer error, even though all three important words appear.

Grammatical subject, semantic agent and discourse topic are three separate concepts. In “The samples were tested by the laboratory,” the samples are grammatical subject and patient; the laboratory is a metonymic institutional agent. In “At the laboratory, the samples were tested,” the laboratory is a location; no tester is named. In “The laboratory was tested,” the laboratory itself may be the tested entity. A faithful translation sometimes must reconstruct a full clause in a different order to prevent these roles from drifting.

Frame-specific roles can be more informative than generic labels. For buying, one may need buyer, seller, goods and money. For teaching, teacher, student, subject and medium. For damaging, damager, damaged entity, instrument and cause. For perceiving, experiencer, stimulus and sensory channel. For communicating, speaker, addressee, message and communication medium. The generic agent–patient shorthand is useful for triage but can mislead if it flattens a recipient into an affected patient or treats every experiencer as intentional. Before translation, choose labels that make the relevant distinctions visible.

Passive voice is a useful but imperfect clue. English often promotes a non-agent participant to grammatical subject, but constructions differ across languages and across individual verbs. A source passive may allow the agent to be stated, omitted, backgrounded or treated as irrelevant. A language may have several voice systems or may prefer an impersonal clause. Translating every source passive as a target passive is not a quality rule; it is a formal recipe that may produce unnatural or even misleading prose. Conversely, automatically making passives active may force the translator to name an agent where none is known. Prefer a target construction preserving event structure and discourse priorities.

Causation requires its own field. A person can intentionally perform an action, accidentally cause an outcome, permit another person to act, arrange for someone to act, or merely be present when something happens. “Niko had the roof repaired” normally implies arranging a repair, not personally making it. “Niko let the roof be repaired” may describe permission rather than arrangement. “Niko repaired the roof” attributes the physical action to Niko. These are not stylistic rewrites of the same participant pattern. The causative relation must survive even when a target language uses a lexical verb or multi-clause expression rather than an English have/get causative.

Experiencers further expose why a simple subject-object swap fails. “The announcement worried Priya” and “Priya was worried by the announcement” may be close in propositional content but differ in emphasis and sometimes in the temporal reading of an emotional state. “Priya worried about the announcement” may mean she actively contemplated it over a period rather than immediately reacted. Without context, it is dangerous to present the three sentences as perfectly equivalent. The translator needs a role map and a discourse-context check.

Instruments can also become false agents. “The crane lifted the beam” may be acceptable shorthand for a machine operating under human control, but “The operator lifted the beam with the crane” asserts an operator. In a safety narrative, the distinction matters: was the crane programmed, manually driven, remotely controlled or malfunctioning? Translation must not silently choose an answer. An instrument is not necessarily an independent intentional agent, and a machine’s grammatical subject status does not prove it acted autonomously.

Institutional agents deserve cautious treatment. “The board approved the change” attributes the act to an institution acting through its procedures. “The chair approved the change” singles out an officeholder. “The change was approved” reports an outcome without identifying a decision-maker. A summary that substitutes these forms could distort governance history or individual responsibility. The translator should preserve the level of attribution: organization, group, role, named person, unidentified participant, or no attribution at all.

Automatic methods can help identify candidate grammatical relations, but cannot replace semantic checking. Universal Dependencies documents such grammatical relations as nsubj, obj, nsubj:pass and obl:agent; they are useful cues, not complete logical or factual verdicts. Multidimensional Quality Metrics recognizes accuracy problems including omissions, additions and mistranslations of technical relationships. A responsible process uses these resources to guide review, then asks a human to check what the sentence actually asserts in context.

The eighteen-pass participant-role translation method

Identify the communicative task. Is the text a witness account, novel, subtitle, instruction, abstract, contract, interface message or internal note? The task determines the cost of a wrong attribution.
Freeze the source version. Record the exact source wording, context and revision. A corrected source can invalidate earlier translator assumptions.
Segment by events, not sentences. One sentence can describe several actions, mental states, results and counterfactual claims.
Name each predicate. For every event, ask what happened: transferred, noticed, blocked, damaged, allowed, caused, requested, prevented, remembered or changed.
Identify role-bearing expressions. Map who or what fills each predicate’s essential roles, including implied participants when the language grammatically requires them.
Separate agent from grammatical subject. The subject may be affected, receiving, located, experiencing or simply serving as the sentence’s topic.
Record omissions explicitly. Mark an agent unknown, deliberately suppressed, irrelevant or recoverable from the previous sentence; do not collapse these states.
Test volition. Distinguish willing action from accident, involuntary experience, natural cause and unproved intent.
Test causal depth. Distinguish acting oneself, causing someone else to act, permitting, ordering, enabling and merely observing.
Resolve references cautiously. A pronoun may pick up a participant from earlier discourse. Do not invent a referent when context leaves it open.
Check voice and valency. A source passive, antipassive, applicative, causative or impersonal construction may alter which participants are grammatical core arguments.
Protect information structure. Preserve what is being contrasted or foregrounded, without mistaking grammatical packaging for an additional factual commitment.
Draft natural target language. Choose appropriate local syntax, not forced word order; use explicit agent expressions only when the source licenses them.
Back-map roles. From the target alone, reconstruct its event ledger. Compare it field by field with the source ledger.
Run counterexample tests. Could the source be true but the translation false? Could the translation be true but the source false? Investigate every yes.
Review reader inference. Ask a target-language reader who acted, what changed, whether responsibility is asserted and whether unknown roles remain unknown.
Assign severity and evidence. Distinguish stylistic discomfort from a material change in agency, obligation, identity, safety or accountability.
Record and release. Save the approved target, unresolved uncertainties, reviewer sign-off and any instructions for later source revisions.

This workflow is deliberately staged: first reconstruct meaning, then form a natural target, then test the target independently. Reversing the order often leads the editor to defend an elegant first draft instead of comparing it with the source. No method should encourage the reviewer to manufacture missing source facts merely to make a target sentence sound more explicit.

Five core tests for translation fidelity

1. The same-world test

Imagine one possible world in which the source is true. Would the target necessarily be true as well? If a report says “The seal failed during transport,” a translated claim that “The driver broke the seal” is not guaranteed to be true in the same world: mechanical weakness, vibration or temperature could explain the failure. The translation added an actor, action type and causal path. The same-world test catches this without requiring deep knowledge of formal semantics.

2. The reverse-entailment test

Ask whether a target may be true in a world in which the source is false. “A reviewer signed the record” is more specific than “The record was signed.” A source claiming the former cannot always be reduced to the latter without losing a named responsibility. Conversely, a source claiming only the latter cannot automatically be expanded into the former. These statements may differ not in spelling but in their logical information content.

3. The participant substitution test

Replace the source agent with another plausible agent, keeping the affected entity and outcome constant. If the source would become false but the target remains true, the target may have suppressed agent identity. For example, “The vendor installed the sensor” changes truth value if the installer was the building manager, while “The sensor was installed” can remain true under both histories. This test is helpful for incident records and agreements.

4. The intent reversal test

Keep the action and outcome, but change whether the participant acted deliberately. A cautious description such as “The tray tipped as it was moved” may be compatible with an accident. “The worker deliberately tipped the tray” is not. Ask whether the translated version presupposes intention, knowledge or control that the source has not established.

5. The omitted-agent recovery test

Remove surrounding context and ask a target reader whether the identity of the agent is stated, inferable, or wholly unknown. A translation that substitutes a specific institutional agent for a contextually recoverable pronoun should be checked against the preceding sentence. Where reference remains ambiguous, use a legitimate ambiguity-preserving construction or request clarification; do not conceal the problem behind a seemingly decisive noun.

Failure modes: forty-four ways participant relationships drift

The following cases are not target-language prescriptions. They are deliberately compact source statements, flawed English glosses of possible translations, and editorial remedies. In production, the reviewer must inspect the actual target-language form, its conversational implicatures and its surrounding document. The point of the diagnostic is to expose a shift in participant roles that an apparently grammatical sentence can conceal.

Accuracy should be judged according to the decision the reader must make. A missing named actor in a credit line may be a serious attribution error; a missing actor in a safety instruction may be critical; a missing actor in a purely decorative caption may be minor. Neither passive voice nor active voice has a universal severity rating. Review the consequence, not merely the construction.

1. Unjustified named agent

Source meaning: “The package was opened before inspection.” Flawed target gloss: “The fulfilment clerk opened the package before inspection.” The error is not a cosmetic change in voice: Opening is reported, but the source neither names a clerk nor establishes where the opening occurred. This is the kind of drift that can hide inside a fluent sentence because readers receive a coherent narrative even when it is not the source narrative.

Repair and diagnostic: Keep the opening and inspection sequence; treat the agent as unidentified until reliable context names one. Would the corrected target still be true if a carrier or automatic device opened the package? Ask a second reviewer to reconstruct an actor–event–affected-participant map from the target alone. Then compare that map with the frozen source and its documented context. Reject a version that states a more specific participant, mental state, event completion or attribution than the source warrants; if the source itself is ambiguous, preserve the legitimate uncertainty or ask for clarification.

2. Erased named agent

Source meaning: “The editor removed the paragraph after review.” Flawed target gloss: “The paragraph was removed after review.” The error is not a cosmetic change in voice: The source attributes removal to the editor; the target preserves the action but drops the assigned actor. This is the kind of drift that can hide inside a fluent sentence because readers receive a coherent narrative even when it is not the source narrative.

Repair and diagnostic: Retain the editor’s role, even if the target language naturally puts the paragraph first. Can readers identify the editor, rather than any reviewer, as the remover? Ask a second reviewer to reconstruct an actor–event–affected-participant map from the target alone. Then compare that map with the frozen source and its documented context. Reject a version that states a more specific participant, mental state, event completion or attribution than the source warrants; if the source itself is ambiguous, preserve the legitimate uncertainty or ask for clarification.

3. Instrument mistaken for responsible operator

Source meaning: “The automated gate detected the badge.” Flawed target gloss: “The guard detected the badge at the automated gate.” The error is not a cosmetic change in voice: The original identifies the detection system, not a guard’s personal observation. This is the kind of drift that can hide inside a fluent sentence because readers receive a coherent narrative even when it is not the source narrative.

Repair and diagnostic: Describe the gate’s detection without assuming who monitored it. Could the gate detect the badge with no guard watching the screen? Ask a second reviewer to reconstruct an actor–event–affected-participant map from the target alone. Then compare that map with the frozen source and its documented context. Reject a version that states a more specific participant, mental state, event completion or attribution than the source warrants; if the source itself is ambiguous, preserve the legitimate uncertainty or ask for clarification.

4. Accidental cause made intentional

Source meaning: “During loading, the trolley struck the display.” Flawed target gloss: “The loader deliberately struck the display with the trolley.” The error is not a cosmetic change in voice: Movement and impact are known; human intention, identity and direction are not. This is the kind of drift that can hide inside a fluent sentence because readers receive a coherent narrative even when it is not the source narrative.

Repair and diagnostic: Preserve the impact and loading circumstances, but not an inferred deliberate blow. Is accidental collision still possible under the translation? Ask a second reviewer to reconstruct an actor–event–affected-participant map from the target alone. Then compare that map with the frozen source and its documented context. Reject a version that states a more specific participant, mental state, event completion or attribution than the source warrants; if the source itself is ambiguous, preserve the legitimate uncertainty or ask for clarification.

5. Intentional act disguised as event

Source meaning: “The supervisor disabled the alarm before testing.” Flawed target gloss: “The alarm stopped before testing.” The error is not a cosmetic change in voice: The target removes the supervisor and the disabling action, weakening accountability and chronology. This is the kind of drift that can hide inside a fluent sentence because readers receive a coherent narrative even when it is not the source narrative.

Repair and diagnostic: Preserve who disabled what and when, without assuming why. Would a battery failure satisfy the target but contradict the source? Ask a second reviewer to reconstruct an actor–event–affected-participant map from the target alone. Then compare that map with the frozen source and its documented context. Reject a version that states a more specific participant, mental state, event completion or attribution than the source warrants; if the source itself is ambiguous, preserve the legitimate uncertainty or ask for clarification.

6. State mistaken for event

Source meaning: “The cabinet is locked.” Flawed target gloss: “Someone locked the cabinet.” The error is not a cosmetic change in voice: The source describes a current state, not necessarily a particular locking event. This is the kind of drift that can hide inside a fluent sentence because readers receive a coherent narrative even when it is not the source narrative.

Repair and diagnostic: Use a state predicate where available; do not add an actor or time. Could the cabinet have remained locked since its manufacture? Ask a second reviewer to reconstruct an actor–event–affected-participant map from the target alone. Then compare that map with the frozen source and its documented context. Reject a version that states a more specific participant, mental state, event completion or attribution than the source warrants; if the source itself is ambiguous, preserve the legitimate uncertainty or ask for clarification.

7. Event mistaken for stable state

Source meaning: “The panel locked when power returned.” Flawed target gloss: “The panel was locked.” The error is not a cosmetic change in voice: The target drops the transition triggered by returning power. This is the kind of drift that can hide inside a fluent sentence because readers receive a coherent narrative even when it is not the source narrative.

Repair and diagnostic: Keep the transition and its timing, choosing an inchoative or eventive target construction. Would the translation be false if the panel had been unlocked until power returned? Ask a second reviewer to reconstruct an actor–event–affected-participant map from the target alone. Then compare that map with the frozen source and its documented context. Reject a version that states a more specific participant, mental state, event completion or attribution than the source warrants; if the source itself is ambiguous, preserve the legitimate uncertainty or ask for clarification.

8. Experiencer presented as voluntary actor

Source meaning: “Rina heard a faint alarm.” Flawed target gloss: “Rina listened for the alarm.” The error is not a cosmetic change in voice: Hearing does not prove that Rina intentionally attended or searched for the sound. This is the kind of drift that can hide inside a fluent sentence because readers receive a coherent narrative even when it is not the source narrative.

Repair and diagnostic: Retain passive perception; distinguish listen from hear. Can the source be true while Rina is distracted? Ask a second reviewer to reconstruct an actor–event–affected-participant map from the target alone. Then compare that map with the frozen source and its documented context. Reject a version that states a more specific participant, mental state, event completion or attribution than the source warrants; if the source itself is ambiguous, preserve the legitimate uncertainty or ask for clarification.

9. Perception turned into a physical act

Source meaning: “The visitor noticed a scratch.” Flawed target gloss: “The visitor scratched the surface.” The error is not a cosmetic change in voice: A perception predicate has been replaced by an event where the visitor causes damage. This is the kind of drift that can hide inside a fluent sentence because readers receive a coherent narrative even when it is not the source narrative.

Repair and diagnostic: Preserve the visitor as experiencer and the scratch as noticed stimulus. Who created the scratch, and does the original say? Ask a second reviewer to reconstruct an actor–event–affected-participant map from the target alone. Then compare that map with the frozen source and its documented context. Reject a version that states a more specific participant, mental state, event completion or attribution than the source warrants; if the source itself is ambiguous, preserve the legitimate uncertainty or ask for clarification.

10. Recipient converted into sender

Source meaning: “Amir received the revised itinerary.” Flawed target gloss: “Amir sent the revised itinerary.” The error is not a cosmetic change in voice: One party’s receipt is incorrectly rendered as transmission initiated by that party. This is the kind of drift that can hide inside a fluent sentence because readers receive a coherent narrative even when it is not the source narrative.

Repair and diagnostic: Mark Amir as recipient, leave sender unspecified. Does the target still show the itinerary moving toward Amir? Ask a second reviewer to reconstruct an actor–event–affected-participant map from the target alone. Then compare that map with the frozen source and its documented context. Reject a version that states a more specific participant, mental state, event completion or attribution than the source warrants; if the source itself is ambiguous, preserve the legitimate uncertainty or ask for clarification.

11. Beneficiary confused with worker

Source meaning: “The mural was restored for the museum.” Flawed target gloss: “The museum restored the mural.” The error is not a cosmetic change in voice: The museum may benefit from the work without being the actor that carried it out. This is the kind of drift that can hide inside a fluent sentence because readers receive a coherent narrative even when it is not the source narrative.

Repair and diagnostic: Distinguish beneficiary from restorer in the target. Would an external conservation team still fit the source? Ask a second reviewer to reconstruct an actor–event–affected-participant map from the target alone. Then compare that map with the frozen source and its documented context. Reject a version that states a more specific participant, mental state, event completion or attribution than the source warrants; if the source itself is ambiguous, preserve the legitimate uncertainty or ask for clarification.

12. Permission turned into obligation

Source meaning: “Visitors may photograph the display.” Flawed target gloss: “Visitors must photograph the display.” The error is not a cosmetic change in voice: The acting participants are unchanged, but the modal relation changes permission into duty. This is the kind of drift that can hide inside a fluent sentence because readers receive a coherent narrative even when it is not the source narrative.

Repair and diagnostic: Retain the permissive force and check if exceptions apply. Would a visitor who takes no photographs violate the target but not the source? Ask a second reviewer to reconstruct an actor–event–affected-participant map from the target alone. Then compare that map with the frozen source and its documented context. Reject a version that states a more specific participant, mental state, event completion or attribution than the source warrants; if the source itself is ambiguous, preserve the legitimate uncertainty or ask for clarification.

13. Permission turned into completed action

Source meaning: “The curator allowed the team to examine the artifact.” Flawed target gloss: “The team examined the artifact.” The error is not a cosmetic change in voice: A permission or authorization does not prove that the permitted act was executed. This is the kind of drift that can hide inside a fluent sentence because readers receive a coherent narrative even when it is not the source narrative.

Repair and diagnostic: Represent the curator as authorizer and the examination as merely permitted. Could the team cancel the examination after receiving permission? Ask a second reviewer to reconstruct an actor–event–affected-participant map from the target alone. Then compare that map with the frozen source and its documented context. Reject a version that states a more specific participant, mental state, event completion or attribution than the source warrants; if the source itself is ambiguous, preserve the legitimate uncertainty or ask for clarification.

14. Request mistaken for action

Source meaning: “The coordinator asked the driver to stop.” Flawed target gloss: “The coordinator stopped the vehicle.” The error is not a cosmetic change in voice: The requester is made into the actor carrying out an action that may not have occurred. This is the kind of drift that can hide inside a fluent sentence because readers receive a coherent narrative even when it is not the source narrative.

Repair and diagnostic: Keep requester, addressee and requested act distinct. Can the driver refuse the request while the source stays true? Ask a second reviewer to reconstruct an actor–event–affected-participant map from the target alone. Then compare that map with the frozen source and its documented context. Reject a version that states a more specific participant, mental state, event completion or attribution than the source warrants; if the source itself is ambiguous, preserve the legitimate uncertainty or ask for clarification.

15. Instruction mistaken for consent

Source meaning: “The officer instructed the witness to sign.” Flawed target gloss: “The witness agreed to sign.” The error is not a cosmetic change in voice: An instruction from one actor is not voluntary agreement by another. This is the kind of drift that can hide inside a fluent sentence because readers receive a coherent narrative even when it is not the source narrative.

Repair and diagnostic: Keep the direction of communication and leave compliance unresolved. Does the original establish that the witness accepted the instruction? Ask a second reviewer to reconstruct an actor–event–affected-participant map from the target alone. Then compare that map with the frozen source and its documented context. Reject a version that states a more specific participant, mental state, event completion or attribution than the source warrants; if the source itself is ambiguous, preserve the legitimate uncertainty or ask for clarification.

16. Assistance mistaken for primary performance

Source meaning: “Luca helped Nia install the shelf.” Flawed target gloss: “Luca installed the shelf.” The error is not a cosmetic change in voice: The helper is promoted into the sole performer, and Nia disappears. This is the kind of drift that can hide inside a fluent sentence because readers receive a coherent narrative even when it is not the source narrative.

Repair and diagnostic: Preserve the collaborative asymmetry; do not erase the primary participant. Can Nia have performed most of the installation? Ask a second reviewer to reconstruct an actor–event–affected-participant map from the target alone. Then compare that map with the frozen source and its documented context. Reject a version that states a more specific participant, mental state, event completion or attribution than the source warrants; if the source itself is ambiguous, preserve the legitimate uncertainty or ask for clarification.

17. Arrangement confused with personal performance

Source meaning: “The owner had the roof inspected.” Flawed target gloss: “The owner inspected the roof.” The error is not a cosmetic change in voice: Causative have normally signals commissioning or arranging, not necessarily performing. This is the kind of drift that can hide inside a fluent sentence because readers receive a coherent narrative even when it is not the source narrative.

Repair and diagnostic: Distinguish organizer from actual inspector unless context establishes identity. Would the source remain true with a separate inspector? Ask a second reviewer to reconstruct an actor–event–affected-participant map from the target alone. Then compare that map with the frozen source and its documented context. Reject a version that states a more specific participant, mental state, event completion or attribution than the source warrants; if the source itself is ambiguous, preserve the legitimate uncertainty or ask for clarification.

18. Indirect cause made direct action

Source meaning: “The manager caused the system to restart.” Flawed target gloss: “The manager restarted the system manually.” The error is not a cosmetic change in voice: The source identifies causal involvement but does not entail manual operation. This is the kind of drift that can hide inside a fluent sentence because readers receive a coherent narrative even when it is not the source narrative.

Repair and diagnostic: Preserve the causal link without inventing physical intervention. Could the manager have changed an automated schedule instead? Ask a second reviewer to reconstruct an actor–event–affected-participant map from the target alone. Then compare that map with the frozen source and its documented context. Reject a version that states a more specific participant, mental state, event completion or attribution than the source warrants; if the source itself is ambiguous, preserve the legitimate uncertainty or ask for clarification.

19. Natural force anthropomorphized

Source meaning: “Strong wind lifted the cover.” Flawed target gloss: “A worker lifted the cover during strong wind.” The error is not a cosmetic change in voice: A natural cause is replaced by a human agent, changing the explanation. This is the kind of drift that can hide inside a fluent sentence because readers receive a coherent narrative even when it is not the source narrative.

Repair and diagnostic: Keep the wind as cause; if target style requires a subject, use an appropriate natural-event construction. Would the event occur without a worker touching the cover? Ask a second reviewer to reconstruct an actor–event–affected-participant map from the target alone. Then compare that map with the frozen source and its documented context. Reject a version that states a more specific participant, mental state, event completion or attribution than the source warrants; if the source itself is ambiguous, preserve the legitimate uncertainty or ask for clarification.

20. Actor reduced to location

Source meaning: “The laboratory analyzed the sample.” Flawed target gloss: “The sample was analyzed at the laboratory.” The error is not a cosmetic change in voice: Institutional performance is downgraded to place of performance. This is the kind of drift that can hide inside a fluent sentence because readers receive a coherent narrative even when it is not the source narrative.

Repair and diagnostic: Retain the laboratory as institutional actor unless only location is warranted. Could an outside contractor analyze the sample inside the laboratory? Ask a second reviewer to reconstruct an actor–event–affected-participant map from the target alone. Then compare that map with the frozen source and its documented context. Reject a version that states a more specific participant, mental state, event completion or attribution than the source warrants; if the source itself is ambiguous, preserve the legitimate uncertainty or ask for clarification.

21. Location turned into institution

Source meaning: “The sample was analyzed at the laboratory.” Flawed target gloss: “The laboratory analyzed the sample.” The error is not a cosmetic change in voice: A location phrase is inflated into an organization-level agency claim. This is the kind of drift that can hide inside a fluent sentence because readers receive a coherent narrative even when it is not the source narrative.

Repair and diagnostic: Translate the place of analysis, not the assumed operator. Could a visiting team carry out the analysis there? Ask a second reviewer to reconstruct an actor–event–affected-participant map from the target alone. Then compare that map with the frozen source and its documented context. Reject a version that states a more specific participant, mental state, event completion or attribution than the source warrants; if the source itself is ambiguous, preserve the legitimate uncertainty or ask for clarification.

22. Board decision attributed to chair

Source meaning: “The committee approved the proposal.” Flawed target gloss: “The chair approved the proposal.” The error is not a cosmetic change in voice: A collective decision is assigned to a particular officer. This is the kind of drift that can hide inside a fluent sentence because readers receive a coherent narrative even when it is not the source narrative.

Repair and diagnostic: Maintain the committee’s collective role or its approved decision. Would the chair have unilateral authority in the source? Ask a second reviewer to reconstruct an actor–event–affected-participant map from the target alone. Then compare that map with the frozen source and its documented context. Reject a version that states a more specific participant, mental state, event completion or attribution than the source warrants; if the source itself is ambiguous, preserve the legitimate uncertainty or ask for clarification.

23. Chair decision attributed to board

Source meaning: “The chair approved the proposal.” Flawed target gloss: “The committee approved the proposal.” The error is not a cosmetic change in voice: The translation changes decision-making authority and institutional record. This is the kind of drift that can hide inside a fluent sentence because readers receive a coherent narrative even when it is not the source narrative.

Repair and diagnostic: Retain the chair as approver, not the whole body. Could a chair approve a preliminary step before a later board vote? Ask a second reviewer to reconstruct an actor–event–affected-participant map from the target alone. Then compare that map with the frozen source and its documented context. Reject a version that states a more specific participant, mental state, event completion or attribution than the source warrants; if the source itself is ambiguous, preserve the legitimate uncertainty or ask for clarification.

24. Colleague voice confused with author

Source meaning: “The message from the director was read aloud by the assistant.” Flawed target gloss: “The assistant announced the new policy.” The error is not a cosmetic change in voice: Reading aloud is not necessarily originating or authorizing the content. This is the kind of drift that can hide inside a fluent sentence because readers receive a coherent narrative even when it is not the source narrative.

Repair and diagnostic: Keep message author, performer and communicative event separate. Could the assistant read the message while disagreeing with its policy? Ask a second reviewer to reconstruct an actor–event–affected-participant map from the target alone. Then compare that map with the frozen source and its documented context. Reject a version that states a more specific participant, mental state, event completion or attribution than the source warrants; if the source itself is ambiguous, preserve the legitimate uncertainty or ask for clarification.

25. Speaker separated from quote

Source meaning: “Nila reported that the valve had been checked.” Flawed target gloss: “Nila checked the valve.” The error is not a cosmetic change in voice: A reporting act is mistaken for the reported underlying action. This is the kind of drift that can hide inside a fluent sentence because readers receive a coherent narrative even when it is not the source narrative.

Repair and diagnostic: Represent Nila as reporter; identify the checker only if the report does. Could the valve have been checked by a different technician? Ask a second reviewer to reconstruct an actor–event–affected-participant map from the target alone. Then compare that map with the frozen source and its documented context. Reject a version that states a more specific participant, mental state, event completion or attribution than the source warrants; if the source itself is ambiguous, preserve the legitimate uncertainty or ask for clarification.

26. Allegation converted into established act

Source meaning: “The witness alleged that the driver ignored the warning.” Flawed target gloss: “The driver ignored the warning.” The error is not a cosmetic change in voice: The translator upgrades an attributed assertion into narrator-endorsed fact. This is the kind of drift that can hide inside a fluent sentence because readers receive a coherent narrative even when it is not the source narrative.

Repair and diagnostic: Keep who alleges what, with attribution and evidential status intact. Would a later investigation disprove the allegation without falsifying the fact of alleging? Ask a second reviewer to reconstruct an actor–event–affected-participant map from the target alone. Then compare that map with the frozen source and its documented context. Reject a version that states a more specific participant, mental state, event completion or attribution than the source warrants; if the source itself is ambiguous, preserve the legitimate uncertainty or ask for clarification.

27. Mediated report turned into observation

Source meaning: “The analyst learned from the log that a key was used.” Flawed target gloss: “The analyst saw someone use a key.” The error is not a cosmetic change in voice: Evidence source changes from record to direct observation, and an unnamed agent is added. This is the kind of drift that can hide inside a fluent sentence because readers receive a coherent narrative even when it is not the source narrative.

Repair and diagnostic: Preserve the analyst’s learning channel and the event known from the log. Could the analyst be away when the key was used? Ask a second reviewer to reconstruct an actor–event–affected-participant map from the target alone. Then compare that map with the frozen source and its documented context. Reject a version that states a more specific participant, mental state, event completion or attribution than the source warrants; if the source itself is ambiguous, preserve the legitimate uncertainty or ask for clarification.

28. Nominalization masks responsible parties

Source meaning: “After review, approval of the plan followed.” Flawed target gloss: “The reviewer approved the plan.” The error is not a cosmetic change in voice: The source offers sequence but does not necessarily identify reviewer and approver as one actor. This is the kind of drift that can hide inside a fluent sentence because readers receive a coherent narrative even when it is not the source narrative.

Repair and diagnostic: Keep the two events and avoid joining their participant identities without evidence. Can separate people review and approve the plan? Ask a second reviewer to reconstruct an actor–event–affected-participant map from the target alone. Then compare that map with the frozen source and its documented context. Reject a version that states a more specific participant, mental state, event completion or attribution than the source warrants; if the source itself is ambiguous, preserve the legitimate uncertainty or ask for clarification.

29. Reflexive role reversal

Source meaning: “The performer injured herself during rehearsal.” Flawed target gloss: “The performer injured a colleague during rehearsal.” The error is not a cosmetic change in voice: A reflexive relationship between agent and patient is lost. This is the kind of drift that can hide inside a fluent sentence because readers receive a coherent narrative even when it is not the source narrative.

Repair and diagnostic: Mark the same participant as both actor and affected person where required. Can the translated sentence describe a second injured person? Ask a second reviewer to reconstruct an actor–event–affected-participant map from the target alone. Then compare that map with the frozen source and its documented context. Reject a version that states a more specific participant, mental state, event completion or attribution than the source warrants; if the source itself is ambiguous, preserve the legitimate uncertainty or ask for clarification.

30. Reciprocal action collapsed

Source meaning: “The neighbors congratulated each other.” Flawed target gloss: “One neighbor congratulated the other.” The error is not a cosmetic change in voice: Reciprocity entails two directional acts, not merely one. This is the kind of drift that can hide inside a fluent sentence because readers receive a coherent narrative even when it is not the source narrative.

Repair and diagnostic: Use a reciprocal form or make both directions explicit. Has each neighbor both given and received congratulations? Ask a second reviewer to reconstruct an actor–event–affected-participant map from the target alone. Then compare that map with the frozen source and its documented context. Reject a version that states a more specific participant, mental state, event completion or attribution than the source warrants; if the source itself is ambiguous, preserve the legitimate uncertainty or ask for clarification.

31. Joint action changed into sequence

Source meaning: “Both engineers lifted the panel together.” Flawed target gloss: “The engineers lifted the panel one after another.” The error is not a cosmetic change in voice: A joint event becomes two sequential events. This is the kind of drift that can hide inside a fluent sentence because readers receive a coherent narrative even when it is not the source narrative.

Repair and diagnostic: Preserve simultaneity and collective participation without fixing individual grip positions. Could only one engineer touch the panel at a time under the source? Ask a second reviewer to reconstruct an actor–event–affected-participant map from the target alone. Then compare that map with the frozen source and its documented context. Reject a version that states a more specific participant, mental state, event completion or attribution than the source warrants; if the source itself is ambiguous, preserve the legitimate uncertainty or ask for clarification.

32. Plural team made singular

Source meaning: “The volunteers carried the crates.” Flawed target gloss: “A volunteer carried the crates.” The error is not a cosmetic change in voice: The target narrows the participating set and may imply a single worker performed all movements. This is the kind of drift that can hide inside a fluent sentence because readers receive a coherent narrative even when it is not the source narrative.

Repair and diagnostic: Respect plural participation without asserting that everyone touched every crate. Could multiple volunteers take different crates? Ask a second reviewer to reconstruct an actor–event–affected-participant map from the target alone. Then compare that map with the frozen source and its documented context. Reject a version that states a more specific participant, mental state, event completion or attribution than the source warrants; if the source itself is ambiguous, preserve the legitimate uncertainty or ask for clarification.

33. Agent changed across coordinated verbs

Source meaning: “Tara opened the file and Sam archived it.” Flawed target gloss: “Tara opened and archived the file.” The error is not a cosmetic change in voice: A second clause’s explicit actor is overwritten by subject sharing. This is the kind of drift that can hide inside a fluent sentence because readers receive a coherent narrative even when it is not the source narrative.

Repair and diagnostic: Track each predicate separately before combining clauses. Would the source allow Sam to archive a file Tara merely opened? Ask a second reviewer to reconstruct an actor–event–affected-participant map from the target alone. Then compare that map with the frozen source and its documented context. Reject a version that states a more specific participant, mental state, event completion or attribution than the source warrants; if the source itself is ambiguous, preserve the legitimate uncertainty or ask for clarification.

34. Controller confused with performer

Source meaning: “The researcher instructed the assistant to calibrate the meter.” Flawed target gloss: “The researcher calibrated the meter.” The error is not a cosmetic change in voice: An instructor’s command is rendered as direct performance. This is the kind of drift that can hide inside a fluent sentence because readers receive a coherent narrative even when it is not the source narrative.

Repair and diagnostic: Keep instruction, addressee and projected task separate. Could the assistant decline or delay calibration? Ask a second reviewer to reconstruct an actor–event–affected-participant map from the target alone. Then compare that map with the frozen source and its documented context. Reject a version that states a more specific participant, mental state, event completion or attribution than the source warrants; if the source itself is ambiguous, preserve the legitimate uncertainty or ask for clarification.

35. Question converted into declaration

Source meaning: “Did the technician replace the sensor?” Flawed target gloss: “The technician replaced the sensor.” The error is not a cosmetic change in voice: An inquiry about an actor-event relation becomes an assertion that the event happened. This is the kind of drift that can hide inside a fluent sentence because readers receive a coherent narrative even when it is not the source narrative.

Repair and diagnostic: Preserve interrogative status and the unresolved truth of the replacement. Would the original remain appropriate if nobody knew the answer? Ask a second reviewer to reconstruct an actor–event–affected-participant map from the target alone. Then compare that map with the frozen source and its documented context. Reject a version that states a more specific participant, mental state, event completion or attribution than the source warrants; if the source itself is ambiguous, preserve the legitimate uncertainty or ask for clarification.

36. Negated event turned into wrong agent

Source meaning: “Maya did not approve the change.” Flawed target gloss: “Maya’s colleague approved the change.” The error is not a cosmetic change in voice: Negating Maya’s approval does not assert any other person’s approval. This is the kind of drift that can hide inside a fluent sentence because readers receive a coherent narrative even when it is not the source narrative.

Repair and diagnostic: Preserve the negated approval relation without inventing another approver. Could the change be unapproved by everyone? Ask a second reviewer to reconstruct an actor–event–affected-participant map from the target alone. Then compare that map with the frozen source and its documented context. Reject a version that states a more specific participant, mental state, event completion or attribution than the source warrants; if the source itself is ambiguous, preserve the legitimate uncertainty or ask for clarification.

37. Ability confused with actual performance

Source meaning: “The robot can carry the crate.” Flawed target gloss: “The robot carried the crate.” The error is not a cosmetic change in voice: Capacity does not entail occurrence; the grammatical actor can be possible rather than actual. This is the kind of drift that can hide inside a fluent sentence because readers receive a coherent narrative even when it is not the source narrative.

Repair and diagnostic: Keep the modal capacity and avoid constructing a completed event. Could the robot be capable yet never switched on? Ask a second reviewer to reconstruct an actor–event–affected-participant map from the target alone. Then compare that map with the frozen source and its documented context. Reject a version that states a more specific participant, mental state, event completion or attribution than the source warrants; if the source itself is ambiguous, preserve the legitimate uncertainty or ask for clarification.

38. Hypothetical action reported as fact

Source meaning: “If the valve fails, the pump will stop.” Flawed target gloss: “The valve failed and the pump stopped.” The error is not a cosmetic change in voice: A conditional future risk is converted into historical outcomes. This is the kind of drift that can hide inside a fluent sentence because readers receive a coherent narrative even when it is not the source narrative.

Repair and diagnostic: Maintain conditional dependency and prospective time. Can the source remain true when the valve never fails? Ask a second reviewer to reconstruct an actor–event–affected-participant map from the target alone. Then compare that map with the frozen source and its documented context. Reject a version that states a more specific participant, mental state, event completion or attribution than the source warrants; if the source itself is ambiguous, preserve the legitimate uncertainty or ask for clarification.

39. Patient confused with property owner

Source meaning: “The tenant’s window was repaired.” Flawed target gloss: “The tenant repaired the window.” The error is not a cosmetic change in voice: A possessive relationship to the affected object is mistaken for responsibility for repair. This is the kind of drift that can hide inside a fluent sentence because readers receive a coherent narrative even when it is not the source narrative.

Repair and diagnostic: Keep tenant–window relationship without making the tenant the worker. Could the building owner arrange the repair? Ask a second reviewer to reconstruct an actor–event–affected-participant map from the target alone. Then compare that map with the frozen source and its documented context. Reject a version that states a more specific participant, mental state, event completion or attribution than the source warrants; if the source itself is ambiguous, preserve the legitimate uncertainty or ask for clarification.

40. Recipient mistaken for target of harm

Source meaning: “The warning was sent to the pilot.” Flawed target gloss: “The pilot was warned for causing the incident.” The error is not a cosmetic change in voice: Receiving a warning is not proof of responsibility for an incident or disciplinary action. This is the kind of drift that can hide inside a fluent sentence because readers receive a coherent narrative even when it is not the source narrative.

Repair and diagnostic: Preserve communication destination and leave causal responsibility open. Does the warning concern a hazard that the pilot did not cause? Ask a second reviewer to reconstruct an actor–event–affected-participant map from the target alone. Then compare that map with the frozen source and its documented context. Reject a version that states a more specific participant, mental state, event completion or attribution than the source warrants; if the source itself is ambiguous, preserve the legitimate uncertainty or ask for clarification.

41. Machine failure blamed on operator

Source meaning: “The sensor failed after recalibration.” Flawed target gloss: “The operator damaged the sensor during recalibration.” The error is not a cosmetic change in voice: Temporal sequence does not entail intentional or accidental personal causation. This is the kind of drift that can hide inside a fluent sentence because readers receive a coherent narrative even when it is not the source narrative.

Repair and diagnostic: Keep failure and chronology without alleging a responsible person. Could an internal defect cause failure after correct calibration? Ask a second reviewer to reconstruct an actor–event–affected-participant map from the target alone. Then compare that map with the frozen source and its documented context. Reject a version that states a more specific participant, mental state, event completion or attribution than the source warrants; if the source itself is ambiguous, preserve the legitimate uncertainty or ask for clarification.

42. Unclear antecedent silently resolved

Source meaning: “When Pat called Lee, they had already left.” Flawed target gloss: “Pat had left before calling Lee.” The error is not a cosmetic change in voice: The original pronoun is potentially ambiguous and does not license a definite choice without context. This is the kind of drift that can hide inside a fluent sentence because readers receive a coherent narrative even when it is not the source narrative.

Repair and diagnostic: Retain supported ambiguity or clarify with the author; record the unresolved reference. Could the source be describing Lee’s departure? Ask a second reviewer to reconstruct an actor–event–affected-participant map from the target alone. Then compare that map with the frozen source and its documented context. Reject a version that states a more specific participant, mental state, event completion or attribution than the source warrants; if the source itself is ambiguous, preserve the legitimate uncertainty or ask for clarification.

43. Headline compresses an allegation

Source meaning: “Report: vendor accused of deleting records.” Flawed target gloss: “Vendor deletes records.” The error is not a cosmetic change in voice: A headline reporting an accusation is flattened into a finding of fact. This is the kind of drift that can hide inside a fluent sentence because readers receive a coherent narrative even when it is not the source narrative.

Repair and diagnostic: Retain the allegation and its reported status even under headline length pressure. Would acquittal or rebuttal change whether the original report was published? Ask a second reviewer to reconstruct an actor–event–affected-participant map from the target alone. Then compare that map with the frozen source and its documented context. Reject a version that states a more specific participant, mental state, event completion or attribution than the source warrants; if the source itself is ambiguous, preserve the legitimate uncertainty or ask for clarification.

44. Subtitle loses second actor

Source meaning: “I asked her to leave, but she refused.” Flawed target gloss: “I left.” The error is not a cosmetic change in voice: Compression erases both requester and addressee, reverses event outcome, and destroys the refusal. This is the kind of drift that can hide inside a fluent sentence because readers receive a coherent narrative even when it is not the source narrative.

Repair and diagnostic: Keep the two speech-act roles and the refusal, shortening less important connective detail first. Can a viewer identify who asked whom and whether anyone departed? Ask a second reviewer to reconstruct an actor–event–affected-participant map from the target alone. Then compare that map with the frozen source and its documented context. Reject a version that states a more specific participant, mental state, event completion or attribution than the source warrants; if the source itself is ambiguous, preserve the legitimate uncertainty or ask for clarification.

How the failure catalogue should be used

A catalogue is not a replacement for context. Begin with event parsing, then classify a candidate discrepancy as addition, omission, wrong relation, wrong agency, wrong volition, wrong evidential status, or misleading information structure. Some examples implicate two categories at once; an omitted reporter can also create an invented factual certainty. Keep the rationale attached to the specific source span. Where the target is idiomatic but the reader draws a different factual inference, prefer explicit neutral wording over surface elegance. Conversely, where a target syntax requires an overt grammatical placeholder, do not confuse that placeholder with an asserted real-world actor.

Worked diagnostics: original event-role tests across eight practical domains

Each scenario starts with a hypothetical English source meaning, then a misleading target-language gloss expressed in English and a safer meaning-preserving gloss. These are not literal translations for a specified language pair. The pedagogical work happens in the role comparison and the independent counterexample question. This avoids teaching unverified foreign-language forms while giving multilingual professionals a repeatable acceptance test.

How to read the diagnostic entries

A good target-language rendering may use a different clause order, voice, case pattern, auxiliary, aspect marker or idiom. Mark a candidate as correct only when a competent target reader can reconstruct the source event, its participant roles and its explicit unknowns. A gloss is a meaning sketch, not a promise that any given target language uses the same wording.

Diagnostic 1: Customer support — participant relationship

Source: “The refund was issued after the duplicate charge was confirmed.” Misleading gloss: “The support agent issued the refund personally.” Meaning-preserving gloss: “The refund was issued after confirmation of a duplicate charge.” These are deliberately competing meanings, not rival stylistic preferences. Issuer identity is left unspecified; confirmation is a condition, not a named decision-maker. A reviewer should not accept the misleading version merely because the document still sounds coherent.

Diagnostic question: Can a billing system or another department issue the refund without contradicting the source? Use the question to construct an alternative possible history. If the source is true in that history while the misleading gloss is false, the two sentences are not semantically equivalent. The repaired version must retain the role allocation and any relevant difference between observation, communication, command, permission, capability, completed action and resulting state.

Translation and release decision: Draft in the target language without forcing English syntax. Then annotate that draft from the target side: what event does it assert, who acts, who receives, who is affected, what remains unresolved, and which inference is only suggested? Compare each answer with the source. If the target morphology requires an explicit subject or participant marker, choose a form that does not introduce an unsupported real-world actor. Record the finding in the reviewer log and verify the full paragraph, not just the isolated sentence.

Diagnostic 2: Customer support — participant relationship

Source: “The customer was informed that the replacement could take two weeks.” Misleading gloss: “The customer agreed to wait two weeks.” Meaning-preserving gloss: “The customer was told that replacement might take up to two weeks.” These are deliberately competing meanings, not rival stylistic preferences. Information provided is not assent; the uncertain duration is not a commitment to a full two-week wait. A reviewer should not accept the misleading version merely because the document still sounds coherent.

Diagnostic question: What evidence, if any, shows agreement by the recipient? Use the question to construct an alternative possible history. If the source is true in that history while the misleading gloss is false, the two sentences are not semantically equivalent. The repaired version must retain the role allocation and any relevant difference between observation, communication, command, permission, capability, completed action and resulting state.

Diagnostic 3: Customer support — participant relationship

Source: “A service ticket was opened automatically when the error recurred.” Misleading gloss: “An employee opened a ticket because the error recurred.” Meaning-preserving gloss: “The system automatically opened a service ticket after the error recurred.” These are deliberately competing meanings, not rival stylistic preferences. Automatic triggering is material; adding a human makes the workflow and audit record less accurate. A reviewer should not accept the misleading version merely because the document still sounds coherent.

Diagnostic question: Could the ticket exist outside business hours with no employee logged in? Use the question to construct an alternative possible history. If the source is true in that history while the misleading gloss is false, the two sentences are not semantically equivalent. The repaired version must retain the role allocation and any relevant difference between observation, communication, command, permission, capability, completed action and resulting state.

Diagnostic 4: Customer support — participant relationship

Source: “The courier left the parcel with the concierge.” Misleading gloss: “The concierge delivered the parcel to the courier.” Meaning-preserving gloss: “The courier handed the parcel to the concierge.” These are deliberately competing meanings, not rival stylistic preferences. The courier is the transfer agent, concierge the recipient and parcel the theme; direction matters more than word order. A reviewer should not accept the misleading version merely because the document still sounds coherent.

Diagnostic question: Which person ends with possession after the handover? Use the question to construct an alternative possible history. If the source is true in that history while the misleading gloss is false, the two sentences are not semantically equivalent. The repaired version must retain the role allocation and any relevant difference between observation, communication, command, permission, capability, completed action and resulting state.

Diagnostic 5: Customer support — participant relationship

Source: “The account holder requested cancellation, but the plan remained active.” Misleading gloss: “The account holder cancelled the plan.” Meaning-preserving gloss: “A cancellation request was submitted, but the plan remained active.” These are deliberately competing meanings, not rival stylistic preferences. The requested state change is not the achieved state; the second clause explicitly blocks treating the account as cancelled. A reviewer should not accept the misleading version merely because the document still sounds coherent.

Diagnostic question: Would a billing charge after the request be consistent with the source? Use the question to construct an alternative possible history. If the source is true in that history while the misleading gloss is false, the two sentences are not semantically equivalent. The repaired version must retain the role allocation and any relevant difference between observation, communication, command, permission, capability, completed action and resulting state.

Diagnostic 6: Customer support — participant relationship

Source: “The user reset the password after the system sent a recovery link.” Misleading gloss: “The system reset the password for the user.” Meaning-preserving gloss: “The system supplied a recovery link, and the user then reset the password.” These are deliberately competing meanings, not rival stylistic preferences. Two sequential actors perform separate operations; merging them obscures control and the security sequence. A reviewer should not accept the misleading version merely because the document still sounds coherent.

Diagnostic question: Who performed the reset, and who only supplied the link? Use the question to construct an alternative possible history. If the source is true in that history while the misleading gloss is false, the two sentences are not semantically equivalent. The repaired version must retain the role allocation and any relevant difference between observation, communication, command, permission, capability, completed action and resulting state.

Diagnostic 7: Customer support — participant relationship

Source: “The package was returned to sender by the carrier.” Misleading gloss: “The sender returned the package.” Meaning-preserving gloss: “The carrier returned the package to its sender.” These are deliberately competing meanings, not rival stylistic preferences. Sender is the destination or recipient of the return, not the return agent. A reviewer should not accept the misleading version merely because the document still sounds coherent.

Diagnostic question: At the end, did the parcel move toward the original sender or away from them? Use the question to construct an alternative possible history. If the source is true in that history while the misleading gloss is false, the two sentences are not semantically equivalent. The repaired version must retain the role allocation and any relevant difference between observation, communication, command, permission, capability, completed action and resulting state.

Diagnostic 8: Customer support — participant relationship

Source: “The seller authorized the repair; a contractor carried it out.” Misleading gloss: “The seller carried out the repair.” Meaning-preserving gloss: “The seller authorized the repair, and a contractor performed the work.” These are deliberately competing meanings, not rival stylistic preferences. Authorization and execution are separate institutional and practical responsibilities. A reviewer should not accept the misleading version merely because the document still sounds coherent.

Diagnostic question: If the repair is faulty, does the text accurately record who performed it? Use the question to construct an alternative possible history. If the source is true in that history while the misleading gloss is false, the two sentences are not semantically equivalent. The repaired version must retain the role allocation and any relevant difference between observation, communication, command, permission, capability, completed action and resulting state.

Diagnostic 9: Customer support — participant relationship

Source: “No technician was assigned when the alarm first appeared.” Misleading gloss: “The technician failed to respond to the alarm.” Meaning-preserving gloss: “At the time the alarm first appeared, no technician had been assigned.” These are deliberately competing meanings, not rival stylistic preferences. Absent assignment cannot be equated with individual failure by an identified technician. A reviewer should not accept the misleading version merely because the document still sounds coherent.

Diagnostic question: Can a particular technician be criticized from this source alone? Use the question to construct an alternative possible history. If the source is true in that history while the misleading gloss is false, the two sentences are not semantically equivalent. The repaired version must retain the role allocation and any relevant difference between observation, communication, command, permission, capability, completed action and resulting state.

Diagnostic 10: Scientific report — participant relationship

Source: “The dye spread through the membrane without agitation.” Misleading gloss: “The researcher spread the dye through the membrane.” Meaning-preserving gloss: “The dye diffused through the membrane without being stirred.” These are deliberately competing meanings, not rival stylistic preferences. Physical process and absence of stirring are known; a deliberate human spreading action is not. A reviewer should not accept the misleading version merely because the document still sounds coherent.

Diagnostic question: Could the described movement occur without a researcher touching the liquid? Use the question to construct an alternative possible history. If the source is true in that history while the misleading gloss is false, the two sentences are not semantically equivalent. The repaired version must retain the role allocation and any relevant difference between observation, communication, command, permission, capability, completed action and resulting state.

Diagnostic 11: Scientific report — participant relationship

Source: “The samples were warmed before the readings were taken.” Misleading gloss: “The analyst warmed and measured each sample.” Meaning-preserving gloss: “The samples were warmed before readings were obtained.” These are deliberately competing meanings, not rival stylistic preferences. The source states two event types and order but does not establish that the same person performed both. A reviewer should not accept the misleading version merely because the document still sounds coherent.

Diagnostic question: Does the report identify anyone who operated the warmer or reader? Use the question to construct an alternative possible history. If the source is true in that history while the misleading gloss is false, the two sentences are not semantically equivalent. The repaired version must retain the role allocation and any relevant difference between observation, communication, command, permission, capability, completed action and resulting state.

Diagnostic 12: Scientific report — participant relationship

Source: “A higher temperature caused the material to expand.” Misleading gloss: “The engineer expanded the material by increasing temperature.” Meaning-preserving gloss: “Heating caused the material to expand.” These are deliberately competing meanings, not rival stylistic preferences. The source reports a physical causal relation, not necessarily an intentional intervention or an engineer. A reviewer should not accept the misleading version merely because the document still sounds coherent.

Diagnostic question: Could ambient temperature variation produce the reported expansion? Use the question to construct an alternative possible history. If the source is true in that history while the misleading gloss is false, the two sentences are not semantically equivalent. The repaired version must retain the role allocation and any relevant difference between observation, communication, command, permission, capability, completed action and resulting state.

Diagnostic 13: Scientific report — participant relationship

Source: “The observer recorded the flashes while another team member timed the intervals.” Misleading gloss: “The observer recorded and timed the flashes.” Meaning-preserving gloss: “One person recorded the flashes while another timed their intervals.” These are deliberately competing meanings, not rival stylistic preferences. Separating the recording role from the timing role may be essential for reproducing the procedure. A reviewer should not accept the misleading version merely because the document still sounds coherent.

Diagnostic question: How many operators are established by the sentence, and what task belongs to each? Use the question to construct an alternative possible history. If the source is true in that history while the misleading gloss is false, the two sentences are not semantically equivalent. The repaired version must retain the role allocation and any relevant difference between observation, communication, command, permission, capability, completed action and resulting state.

Diagnostic 14: Scientific report — participant relationship

Source: “The seedlings were damaged by frost after irrigation.” Misleading gloss: “The irrigation team damaged the seedlings.” Meaning-preserving gloss: “The seedlings suffered frost damage following irrigation.” These are deliberately competing meanings, not rival stylistic preferences. Frost is identified as the immediate damaging cause; the time sequence does not establish negligence by a team. A reviewer should not accept the misleading version merely because the document still sounds coherent.

Diagnostic question: Could careful irrigation precede a damaging overnight freeze? Use the question to construct an alternative possible history. If the source is true in that history while the misleading gloss is false, the two sentences are not semantically equivalent. The repaired version must retain the role allocation and any relevant difference between observation, communication, command, permission, capability, completed action and resulting state.

Diagnostic 15: Scientific report — participant relationship

Source: “The reagent was added after the mixture changed color.” Misleading gloss: “The reagent changed the color of the mixture.” Meaning-preserving gloss: “The mixture changed color before the reagent was added.” These are deliberately competing meanings, not rival stylistic preferences. Reversing the chronology creates a nonexistent causal relation. A reviewer should not accept the misleading version merely because the document still sounds coherent.

Diagnostic question: Could the color change be observed before any reagent addition? Use the question to construct an alternative possible history. If the source is true in that history while the misleading gloss is false, the two sentences are not semantically equivalent. The repaired version must retain the role allocation and any relevant difference between observation, communication, command, permission, capability, completed action and resulting state.

Diagnostic 16: Scientific report — participant relationship

Source: “The device signaled the operator when the filter became clogged.” Misleading gloss: “The operator signaled the device that the filter was clogged.” Meaning-preserving gloss: “The device alerted the operator when its filter became clogged.” These are deliberately competing meanings, not rival stylistic preferences. The alert’s origin, recipient and condition are all reversed in the bad version. A reviewer should not accept the misleading version merely because the document still sounds coherent.

Diagnostic question: Which party sends the signal and which receives it? Use the question to construct an alternative possible history. If the source is true in that history while the misleading gloss is false, the two sentences are not semantically equivalent. The repaired version must retain the role allocation and any relevant difference between observation, communication, command, permission, capability, completed action and resulting state.

Diagnostic 17: Scientific report — participant relationship

Source: “The participants reported discomfort during the simulation.” Misleading gloss: “The researchers observed discomfort during the simulation.” Meaning-preserving gloss: “Participants self-reported discomfort during the simulation.” These are deliberately competing meanings, not rival stylistic preferences. Reported subjective experience is not identical to a researcher’s independent observation. A reviewer should not accept the misleading version merely because the document still sounds coherent.

Diagnostic question: Does the data source consist of participant statements or physical measurements? Use the question to construct an alternative possible history. If the source is true in that history while the misleading gloss is false, the two sentences are not semantically equivalent. The repaired version must retain the role allocation and any relevant difference between observation, communication, command, permission, capability, completed action and resulting state.

Diagnostic 18: Scientific report — participant relationship

Source: “No causal role was established for the additive.” Misleading gloss: “The additive did not cause the change.” Meaning-preserving gloss: “The study did not establish a causal role for the additive.” These are deliberately competing meanings, not rival stylistic preferences. Failure to establish causation is not positive evidence of no causal effect; evidential status is separate from the participant role. A reviewer should not accept the misleading version merely because the document still sounds coherent.

Diagnostic question: Would an inconclusive study satisfy the source but not the flawed gloss? Use the question to construct an alternative possible history. If the source is true in that history while the misleading gloss is false, the two sentences are not semantically equivalent. The repaired version must retain the role allocation and any relevant difference between observation, communication, command, permission, capability, completed action and resulting state.

Diagnostic 19: Governance — participant relationship

Source: “The council asked the secretariat to circulate the draft.” Misleading gloss: “The council circulated the draft.” Meaning-preserving gloss: “The council instructed the secretariat to circulate the draft.” These are deliberately competing meanings, not rival stylistic preferences. Requester and proposed performer are distinct; the source does not assert that circulation happened. A reviewer should not accept the misleading version merely because the document still sounds coherent.

Diagnostic question: Could the secretariat delay the circulation without making the source false? Use the question to construct an alternative possible history. If the source is true in that history while the misleading gloss is false, the two sentences are not semantically equivalent. The repaired version must retain the role allocation and any relevant difference between observation, communication, command, permission, capability, completed action and resulting state.

Diagnostic 20: Governance — participant relationship

Source: “The chair recorded the decision, which the committee had taken earlier.” Misleading gloss: “The chair made the decision.” Meaning-preserving gloss: “The chair documented the committee’s earlier decision.” These are deliberately competing meanings, not rival stylistic preferences. Recordkeeper and decision-maker are different roles, and the earlier time matters. A reviewer should not accept the misleading version merely because the document still sounds coherent.

Diagnostic question: Who exercised judgment and who only formalized the record? Use the question to construct an alternative possible history. If the source is true in that history while the misleading gloss is false, the two sentences are not semantically equivalent. The repaired version must retain the role allocation and any relevant difference between observation, communication, command, permission, capability, completed action and resulting state.

Diagnostic 21: Governance — participant relationship

Source: “The proposal was discussed but not adopted.” Misleading gloss: “The committee rejected the proposal.” Meaning-preserving gloss: “The proposal was discussed and was not adopted.” These are deliberately competing meanings, not rival stylistic preferences. Non-adoption does not necessarily entail rejection, a vote, or any specific decision-maker. A reviewer should not accept the misleading version merely because the document still sounds coherent.

Diagnostic question: Could the meeting simply run out of time? Use the question to construct an alternative possible history. If the source is true in that history while the misleading gloss is false, the two sentences are not semantically equivalent. The repaired version must retain the role allocation and any relevant difference between observation, communication, command, permission, capability, completed action and resulting state.

Diagnostic 22: Governance — participant relationship

Source: “A mediator conveyed the parties’ revised offer.” Misleading gloss: “The mediator proposed the revised offer.” Meaning-preserving gloss: “The mediator passed on an offer made by the parties.” These are deliberately competing meanings, not rival stylistic preferences. Transmission of an offer must not become authorship or endorsement by the mediator. A reviewer should not accept the misleading version merely because the document still sounds coherent.

Diagnostic question: Could the mediator be neutral and personally oppose the offer? Use the question to construct an alternative possible history. If the source is true in that history while the misleading gloss is false, the two sentences are not semantically equivalent. The repaired version must retain the role allocation and any relevant difference between observation, communication, command, permission, capability, completed action and resulting state.

Diagnostic 23: Governance — participant relationship

Source: “The report was submitted on behalf of the working group.” Misleading gloss: “The working group wrote every section of the report.” Meaning-preserving gloss: “The report was submitted for the working group.” These are deliberately competing meanings, not rival stylistic preferences. Representation in submission does not establish who drafted each portion. A reviewer should not accept the misleading version merely because the document still sounds coherent.

Diagnostic question: Could an external editor have prepared the text? Use the question to construct an alternative possible history. If the source is true in that history while the misleading gloss is false, the two sentences are not semantically equivalent. The repaired version must retain the role allocation and any relevant difference between observation, communication, command, permission, capability, completed action and resulting state.

Diagnostic 24: Governance — participant relationship

Source: “The regulator found the application incomplete.” Misleading gloss: “The applicant admitted that the application was incomplete.” Meaning-preserving gloss: “The regulator determined that the application lacked required material.” These are deliberately competing meanings, not rival stylistic preferences. An official assessment cannot be recast as the applicant’s self-admission. A reviewer should not accept the misleading version merely because the document still sounds coherent.

Diagnostic question: What party made the judgment, and was any acceptance recorded? Use the question to construct an alternative possible history. If the source is true in that history while the misleading gloss is false, the two sentences are not semantically equivalent. The repaired version must retain the role allocation and any relevant difference between observation, communication, command, permission, capability, completed action and resulting state.

Diagnostic 25: Governance — participant relationship

Source: “The witness was asked to clarify the sequence.” Misleading gloss: “The witness clarified the sequence.” Meaning-preserving gloss: “The witness received a request for clarification.” These are deliberately competing meanings, not rival stylistic preferences. A request does not entail an answer; the response is an open future event. A reviewer should not accept the misleading version merely because the document still sounds coherent.

Diagnostic question: Could the witness decline to answer? Use the question to construct an alternative possible history. If the source is true in that history while the misleading gloss is false, the two sentences are not semantically equivalent. The repaired version must retain the role allocation and any relevant difference between observation, communication, command, permission, capability, completed action and resulting state.

Diagnostic 26: Governance — participant relationship

Source: “The panel heard objections before announcing its decision.” Misleading gloss: “The panel accepted the objections before announcing its decision.” Meaning-preserving gloss: “The panel listened to objections and then announced a decision.” These are deliberately competing meanings, not rival stylistic preferences. Hearing a submission and agreeing with it are different institutional acts. A reviewer should not accept the misleading version merely because the document still sounds coherent.

Diagnostic question: Would a decision rejecting every objection be consistent with the source? Use the question to construct an alternative possible history. If the source is true in that history while the misleading gloss is false, the two sentences are not semantically equivalent. The repaired version must retain the role allocation and any relevant difference between observation, communication, command, permission, capability, completed action and resulting state.

Diagnostic 27: Governance — participant relationship

Source: “The amendment was drafted by counsel and approved by the board.” Misleading gloss: “The board drafted and approved the amendment.” Meaning-preserving gloss: “Counsel drafted the amendment; the board approved it.” These are deliberately competing meanings, not rival stylistic preferences. Drafting and approving are separate predicates with separate institutional actors. A reviewer should not accept the misleading version merely because the document still sounds coherent.

Diagnostic question: Who should receive credit for authorship and who bears approval authority? Use the question to construct an alternative possible history. If the source is true in that history while the misleading gloss is false, the two sentences are not semantically equivalent. The repaired version must retain the role allocation and any relevant difference between observation, communication, command, permission, capability, completed action and resulting state.

Diagnostic 28: Care communication — participant relationship

Source: “The patient reported dizziness after standing.” Misleading gloss: “The clinician diagnosed dizziness on standing.” Meaning-preserving gloss: “The patient said they felt dizzy after standing.” These are deliberately competing meanings, not rival stylistic preferences. A patient self-report does not equal a clinician’s diagnostic conclusion. A reviewer should not accept the misleading version merely because the document still sounds coherent.

Diagnostic question: Does the source supply a patient’s account or an independently verified finding? Use the question to construct an alternative possible history. If the source is true in that history while the misleading gloss is false, the two sentences are not semantically equivalent. The repaired version must retain the role allocation and any relevant difference between observation, communication, command, permission, capability, completed action and resulting state.

Diagnostic 29: Care communication — participant relationship

Source: “The nurse observed swelling; the patient had not noticed it.” Misleading gloss: “The patient complained of swelling to the nurse.” Meaning-preserving gloss: “The nurse noticed swelling that the patient had not noticed.” These are deliberately competing meanings, not rival stylistic preferences. Source of observation is explicitly the nurse; turning it into a complaint erases a contrast. A reviewer should not accept the misleading version merely because the document still sounds coherent.

Diagnostic question: Could the patient be surprised on hearing the observation? Use the question to construct an alternative possible history. If the source is true in that history while the misleading gloss is false, the two sentences are not semantically equivalent. The repaired version must retain the role allocation and any relevant difference between observation, communication, command, permission, capability, completed action and resulting state.

Diagnostic 30: Care communication — participant relationship

Source: “The device delivered a scheduled reminder to the caregiver.” Misleading gloss: “The caregiver reminded the device of its schedule.” Meaning-preserving gloss: “An automated reminder was sent to the caregiver by the device.” These are deliberately competing meanings, not rival stylistic preferences. The device is a system source of information and the caregiver its recipient. A reviewer should not accept the misleading version merely because the document still sounds coherent.

Diagnostic question: Who receives the prompt, and who or what sends it? Use the question to construct an alternative possible history. If the source is true in that history while the misleading gloss is false, the two sentences are not semantically equivalent. The repaired version must retain the role allocation and any relevant difference between observation, communication, command, permission, capability, completed action and resulting state.

Diagnostic 31: Care communication — participant relationship

Source: “The consultant recommended the procedure, but the person declined it.” Misleading gloss: “The consultant performed the procedure despite refusal.” Meaning-preserving gloss: “The consultant recommended the procedure; the person declined.” These are deliberately competing meanings, not rival stylistic preferences. A recommendation and refusal describe a choice not an executed procedure. A reviewer should not accept the misleading version merely because the document still sounds coherent.

Diagnostic question: Could no procedure take place under the source? Use the question to construct an alternative possible history. If the source is true in that history while the misleading gloss is false, the two sentences are not semantically equivalent. The repaired version must retain the role allocation and any relevant difference between observation, communication, command, permission, capability, completed action and resulting state.

Diagnostic 32: Care communication — participant relationship

Source: “The medicine was withheld because the identity check was incomplete.” Misleading gloss: “The clinician forgot to give the medicine.” Meaning-preserving gloss: “Administration was held pending completion of the identity check.” These are deliberately competing meanings, not rival stylistic preferences. A procedural withholding is not established negligence or forgetfulness; preserve the reason and resulting non-administration. A reviewer should not accept the misleading version merely because the document still sounds coherent.

Diagnostic question: Could the withholding be fully intentional and consistent with policy? Use the question to construct an alternative possible history. If the source is true in that history while the misleading gloss is false, the two sentences are not semantically equivalent. The repaired version must retain the role allocation and any relevant difference between observation, communication, command, permission, capability, completed action and resulting state.

Diagnostic 33: Care communication — participant relationship

Source: “The person could not lift the arm without assistance.” Misleading gloss: “The person refused to lift the arm.” Meaning-preserving gloss: “The person was unable to lift the arm unaided.” These are deliberately competing meanings, not rival stylistic preferences. Inability and unwillingness differ in agency; the source asserts a functional limitation. A reviewer should not accept the misleading version merely because the document still sounds coherent.

Diagnostic question: Could the person be trying very hard to move it? Use the question to construct an alternative possible history. If the source is true in that history while the misleading gloss is false, the two sentences are not semantically equivalent. The repaired version must retain the role allocation and any relevant difference between observation, communication, command, permission, capability, completed action and resulting state.

Diagnostic 34: Care communication — participant relationship

Source: “The visitor was guided to the waiting room by a volunteer.” Misleading gloss: “The visitor directed the volunteer to the waiting room.” Meaning-preserving gloss: “A volunteer guided the visitor to the waiting room.” These are deliberately competing meanings, not rival stylistic preferences. Agent, assisted person and destination are reversed in the flawed gloss. A reviewer should not accept the misleading version merely because the document still sounds coherent.

Diagnostic question: Who knows the route and who is being accompanied? Use the question to construct an alternative possible history. If the source is true in that history while the misleading gloss is false, the two sentences are not semantically equivalent. The repaired version must retain the role allocation and any relevant difference between observation, communication, command, permission, capability, completed action and resulting state.

Diagnostic 35: Care communication — participant relationship

Source: “The clinician explained the available options; no decision was recorded.” Misleading gloss: “The patient chose one of the options after explanation.” Meaning-preserving gloss: “The clinician explained the options, but no decision was documented.” These are deliberately competing meanings, not rival stylistic preferences. Explanation is an information event and choice is a separate possible event; lack of record is not proof of action. A reviewer should not accept the misleading version merely because the document still sounds coherent.

Diagnostic question: Is there evidence of selection, or only evidence that options were described? Use the question to construct an alternative possible history. If the source is true in that history while the misleading gloss is false, the two sentences are not semantically equivalent. The repaired version must retain the role allocation and any relevant difference between observation, communication, command, permission, capability, completed action and resulting state.

Diagnostic 36: Care communication — participant relationship

Source: “The relative asked for an update; the care team had not yet responded.” Misleading gloss: “The care team updated the relative.” Meaning-preserving gloss: “The relative requested an update and had received no response yet.” These are deliberately competing meanings, not rival stylistic preferences. A request and pending response form an uncompleted communication pair. A reviewer should not accept the misleading version merely because the document still sounds coherent.

Diagnostic question: Could the source be true immediately before anyone replies? Use the question to construct an alternative possible history. If the source is true in that history while the misleading gloss is false, the two sentences are not semantically equivalent. The repaired version must retain the role allocation and any relevant difference between observation, communication, command, permission, capability, completed action and resulting state.

Diagnostic 37: Engineering and operations — source-to-target event integrity

Source: “The inspection robot photographed the damaged cable, and the engineer reviewed the images.” Misleading gloss: “The engineer photographed the damaged cable.” Meaning-preserving gloss: “The robot captured the images; the engineer later reviewed them.” Image acquisition and interpretation occur in separate stages, with separate kinds of participants. The revised gloss illustrates the factual distinction a translator needs to preserve; it need not be the best grammatical surface form in any particular target language.

Counterexample probe: Can the engineer conduct an entirely remote review after the robot finishes its route? Model a plausible world in which the source remains true but the flawed gloss would be false. The resulting contrast shows exactly which agent, patient, recipient, experiencer, source of evidence, action or temporal link the poor rendering changed. This is a stronger test than simply asking whether the sentence sounds natural.

Application: Mark the predicate and every participant in the source, including roles that are deliberately left unknown. Check whether the target transforms a communication into a completed act, an observation into an intervention, or a cause into a voluntary human decision. A reviewer must also test how the whole passage frames the participant: an omitted agent may be recoverable from context, while an explicit agent may be crucial for identifying the relevant performer or accountable office. Repair the specific semantic divergence and keep the surrounding tone natural.

Diagnostic 38: Engineering and operations — source-to-target event integrity

Source: “The pressure regulator reduced the flow when the downstream sensor detected a surge.” Misleading gloss: “The downstream sensor reduced the flow.” Meaning-preserving gloss: “The sensor detected a surge; the regulator then reduced flow.” A detector can trigger a controller without physically performing the regulated action. The revised gloss illustrates the factual distinction a translator needs to preserve; it need not be the best grammatical surface form in any particular target language.

Counterexample probe: Which component senses and which component changes the flow? Model a plausible world in which the source remains true but the flawed gloss would be false. The resulting contrast shows exactly which agent, patient, recipient, experiencer, source of evidence, action or temporal link the poor rendering changed. This is a stronger test than simply asking whether the sentence sounds natural.

Diagnostic 39: Engineering and operations — source-to-target event integrity

Source: “The housing cracked during vibration testing.” Misleading gloss: “The tester cracked the housing.” Meaning-preserving gloss: “The housing developed a crack while vibration testing was under way.” The event may be a material failure; the actor and negligence inference are not specified. The revised gloss illustrates the factual distinction a translator needs to preserve; it need not be the best grammatical surface form in any particular target language.

Counterexample probe: Would the source be accurate if the test followed all prescribed procedures? Model a plausible world in which the source remains true but the flawed gloss would be false. The resulting contrast shows exactly which agent, patient, recipient, experiencer, source of evidence, action or temporal link the poor rendering changed. This is a stronger test than simply asking whether the sentence sounds natural.

Diagnostic 40: Engineering and operations — source-to-target event integrity

Source: “An operator requested shutdown; the controller did not execute it.” Misleading gloss: “The operator shut down the controller.” Meaning-preserving gloss: “The operator sent a shutdown request that the controller failed to execute.” The request source and execution mechanism are distinct, and the target must preserve failure to carry out the command. The revised gloss illustrates the factual distinction a translator needs to preserve; it need not be the best grammatical surface form in any particular target language.

Counterexample probe: Was the machine actually shut down? Model a plausible world in which the source remains true but the flawed gloss would be false. The resulting contrast shows exactly which agent, patient, recipient, experiencer, source of evidence, action or temporal link the poor rendering changed. This is a stronger test than simply asking whether the sentence sounds natural.

Diagnostic 41: Engineering and operations — source-to-target event integrity

Source: “The technician isolated the circuit before the software flagged a fault.” Misleading gloss: “The software isolated the circuit before the technician found a fault.” Meaning-preserving gloss: “The technician isolated the circuit; afterward the software flagged a fault.” Actors and chronology are both reversed by a too-hasty clause merge. The revised gloss illustrates the factual distinction a translator needs to preserve; it need not be the best grammatical surface form in any particular target language.

Counterexample probe: Which action occurred first and which participant performed it? Model a plausible world in which the source remains true but the flawed gloss would be false. The resulting contrast shows exactly which agent, patient, recipient, experiencer, source of evidence, action or temporal link the poor rendering changed. This is a stronger test than simply asking whether the sentence sounds natural.

Diagnostic 42: Engineering and operations — source-to-target event integrity

Source: “The beam was lifted using the hoist supplied by the subcontractor.” Misleading gloss: “The subcontractor lifted the beam.” Meaning-preserving gloss: “A hoist supplied by the subcontractor was used to lift the beam.” Supplier of equipment is not necessarily operating agent or lift supervisor. The revised gloss illustrates the factual distinction a translator needs to preserve; it need not be the best grammatical surface form in any particular target language.

Counterexample probe: Could a different contractor operate the supplied hoist? Model a plausible world in which the source remains true but the flawed gloss would be false. The resulting contrast shows exactly which agent, patient, recipient, experiencer, source of evidence, action or temporal link the poor rendering changed. This is a stronger test than simply asking whether the sentence sounds natural.

Diagnostic 43: Engineering and operations — source-to-target event integrity

Source: “The contractor supervised installation by the manufacturer’s team.” Misleading gloss: “The contractor installed the equipment.” Meaning-preserving gloss: “The contractor supervised the installation carried out by the manufacturer’s team.” Supervision, installation and contractual identity form different event roles. The revised gloss illustrates the factual distinction a translator needs to preserve; it need not be the best grammatical surface form in any particular target language.

Counterexample probe: Which organization performed hands-on installation? Model a plausible world in which the source remains true but the flawed gloss would be false. The resulting contrast shows exactly which agent, patient, recipient, experiencer, source of evidence, action or temporal link the poor rendering changed. This is a stronger test than simply asking whether the sentence sounds natural.

Diagnostic 44: Engineering and operations — source-to-target event integrity

Source: “The valve closed after the power loss, although no close command was sent.” Misleading gloss: “The control room closed the valve because power was lost.” Meaning-preserving gloss: “The valve closed following the power loss without a close command.” An automatic or fail-safe physical transition is not an operator-issued action. The revised gloss illustrates the factual distinction a translator needs to preserve; it need not be the best grammatical surface form in any particular target language.

Counterexample probe: Could a fail-safe spring close the valve when control power disappears? Model a plausible world in which the source remains true but the flawed gloss would be false. The resulting contrast shows exactly which agent, patient, recipient, experiencer, source of evidence, action or temporal link the poor rendering changed. This is a stronger test than simply asking whether the sentence sounds natural.

Diagnostic 45: Engineering and operations — source-to-target event integrity

Source: “The diagnostic tool reported overheating; later tests did not confirm it.” Misleading gloss: “The tool caused overheating.” Meaning-preserving gloss: “The tool indicated overheating, which subsequent testing did not verify.” Reporting a suspected state does not cause it, and negative follow-up evidence must remain visible. The revised gloss illustrates the factual distinction a translator needs to preserve; it need not be the best grammatical surface form in any particular target language.

Counterexample probe: What could the tool display falsely without any actual overheating? Model a plausible world in which the source remains true but the flawed gloss would be false. The resulting contrast shows exactly which agent, patient, recipient, experiencer, source of evidence, action or temporal link the poor rendering changed. This is a stronger test than simply asking whether the sentence sounds natural.

Diagnostic 46: Historical and archival — source-to-target event integrity

Source: “A letter attributed the fire to lightning, but the cause was not established.” Misleading gloss: “Lightning caused the fire.” Meaning-preserving gloss: “The letter blamed lightning, while the fire’s cause remained unconfirmed.” The archive preserves attribution by a historical speaker rather than establishing natural causation. The revised gloss illustrates the factual distinction a translator needs to preserve; it need not be the best grammatical surface form in any particular target language.

Counterexample probe: Would later evidence of another cause falsify the letter’s existence? Model a plausible world in which the source remains true but the flawed gloss would be false. The resulting contrast shows exactly which agent, patient, recipient, experiencer, source of evidence, action or temporal link the poor rendering changed. This is a stronger test than simply asking whether the sentence sounds natural.

Diagnostic 47: Historical and archival — source-to-target event integrity

Source: “The chronicle says the bridge was rebuilt during the next reign.” Misleading gloss: “The new ruler rebuilt the bridge.” Meaning-preserving gloss: “According to the chronicle, the bridge was rebuilt in the next reign.” A temporal political reference does not identify the person financing or performing construction. The revised gloss illustrates the factual distinction a translator needs to preserve; it need not be the best grammatical surface form in any particular target language.

Counterexample probe: Could residents or a local authority have organized reconstruction? Model a plausible world in which the source remains true but the flawed gloss would be false. The resulting contrast shows exactly which agent, patient, recipient, experiencer, source of evidence, action or temporal link the poor rendering changed. This is a stronger test than simply asking whether the sentence sounds natural.

Diagnostic 48: Historical and archival — source-to-target event integrity

Source: “The commander ordered the retreat, which several units refused.” Misleading gloss: “The commander retreated with all units.” Meaning-preserving gloss: “The commander issued a retreat order, but several units did not obey it.” A command is not identical to the subordinate action; refusal is an essential result. The revised gloss illustrates the factual distinction a translator needs to preserve; it need not be the best grammatical surface form in any particular target language.

Counterexample probe: Can some units remain in place under the source? Model a plausible world in which the source remains true but the flawed gloss would be false. The resulting contrast shows exactly which agent, patient, recipient, experiencer, source of evidence, action or temporal link the poor rendering changed. This is a stronger test than simply asking whether the sentence sounds natural.

Diagnostic 49: Historical and archival — source-to-target event integrity

Source: “The treaty was signed by delegates acting for their respective councils.” Misleading gloss: “The councils themselves attended and signed the treaty.” Meaning-preserving gloss: “Delegates signed for their councils.” Representation is a legal or institutional relationship, not physical identity between councils and delegates. The revised gloss illustrates the factual distinction a translator needs to preserve; it need not be the best grammatical surface form in any particular target language.

Counterexample probe: Who placed signatures on the document? Model a plausible world in which the source remains true but the flawed gloss would be false. The resulting contrast shows exactly which agent, patient, recipient, experiencer, source of evidence, action or temporal link the poor rendering changed. This is a stronger test than simply asking whether the sentence sounds natural.

Diagnostic 50: Historical and archival — source-to-target event integrity

Source: “A witness remembered seeing smoke, not the moment the roof caught fire.” Misleading gloss: “The witness saw the roof ignite.” Meaning-preserving gloss: “The witness recalled smoke but did not witness the roof’s ignition.” Memory content and direct observation have different evidential boundaries. The revised gloss illustrates the factual distinction a translator needs to preserve; it need not be the best grammatical surface form in any particular target language.

Counterexample probe: Would the account be true if the witness arrived after ignition? Model a plausible world in which the source remains true but the flawed gloss would be false. The resulting contrast shows exactly which agent, patient, recipient, experiencer, source of evidence, action or temporal link the poor rendering changed. This is a stronger test than simply asking whether the sentence sounds natural.

Diagnostic 51: Historical and archival — source-to-target event integrity

Source: “The proclamation was read by a herald and written by an unnamed clerk.” Misleading gloss: “The herald authored the proclamation.” Meaning-preserving gloss: “A herald publicly read a proclamation prepared by an unnamed clerk.” Performance, authorship and role uncertainty must stay apart. The revised gloss illustrates the factual distinction a translator needs to preserve; it need not be the best grammatical surface form in any particular target language.

Counterexample probe: Can the herald speak a document without drafting it? Model a plausible world in which the source remains true but the flawed gloss would be false. The resulting contrast shows exactly which agent, patient, recipient, experiencer, source of evidence, action or temporal link the poor rendering changed. This is a stronger test than simply asking whether the sentence sounds natural.

Diagnostic 52: Historical and archival — source-to-target event integrity

Source: “The record notes that the envoy carried a sealed message.” Misleading gloss: “The envoy wrote the sealed message.” Meaning-preserving gloss: “The envoy transported a sealed message.” Transporting, writing and authorizing a message belong to distinct participants. The revised gloss illustrates the factual distinction a translator needs to preserve; it need not be the best grammatical surface form in any particular target language.

Counterexample probe: Could the envoy be unaware of the message’s contents? Model a plausible world in which the source remains true but the flawed gloss would be false. The resulting contrast shows exactly which agent, patient, recipient, experiencer, source of evidence, action or temporal link the poor rendering changed. This is a stronger test than simply asking whether the sentence sounds natural.

Diagnostic 53: Historical and archival — source-to-target event integrity

Source: “The witnesses were listed but none was recorded as approving the decision.” Misleading gloss: “The witnesses approved the decision.” Meaning-preserving gloss: “The witnesses’ names appeared, but no approval by them was recorded.” Listing people as witnesses does not establish their consent or active participation in the decision. The revised gloss illustrates the factual distinction a translator needs to preserve; it need not be the best grammatical surface form in any particular target language.

Counterexample probe: Is there a signed approval or only a record of presence? Model a plausible world in which the source remains true but the flawed gloss would be false. The resulting contrast shows exactly which agent, patient, recipient, experiencer, source of evidence, action or temporal link the poor rendering changed. This is a stronger test than simply asking whether the sentence sounds natural.

Diagnostic 54: Historical and archival — source-to-target event integrity

Source: “The historian inferred the date from the ledger; the ledger did not state it directly.” Misleading gloss: “The clerk wrote the date in the ledger.” Meaning-preserving gloss: “The date was reconstructed by the historian from ledger evidence.” An inference by a later scholar must not become explicit knowledge recorded by the original clerk. The revised gloss illustrates the factual distinction a translator needs to preserve; it need not be the best grammatical surface form in any particular target language.

Counterexample probe: Could the original ledger omit a date entirely? Model a plausible world in which the source remains true but the flawed gloss would be false. The resulting contrast shows exactly which agent, patient, recipient, experiencer, source of evidence, action or temporal link the poor rendering changed. This is a stronger test than simply asking whether the sentence sounds natural.

Diagnostic 55: Literature and dialogue — source-to-target event integrity

Source: “After the light went out, Ira heard footsteps but saw no one.” Misleading gloss: “Ira followed the person who switched off the light.” Meaning-preserving gloss: “The light went out; Ira heard footsteps but did not see anyone.” The narrator withholds both cause of darkness and identity of a walker; the flawed gloss fabricates a pursuit and actor. The revised gloss illustrates the factual distinction a translator needs to preserve; it need not be the best grammatical surface form in any particular target language.

Counterexample probe: Would the passage still sustain uncertainty about who is nearby? Model a plausible world in which the source remains true but the flawed gloss would be false. The resulting contrast shows exactly which agent, patient, recipient, experiencer, source of evidence, action or temporal link the poor rendering changed. This is a stronger test than simply asking whether the sentence sounds natural.

Diagnostic 56: Literature and dialogue — source-to-target event integrity

Source: “She let the letter fall as the bell rang.” Misleading gloss: “She threw the letter at the sound of the bell.” Meaning-preserving gloss: “She allowed the letter to slip from her hand when the bell rang.” A loss of grip or passive release is not the same as deliberate projection; context may support one, but the isolated source does not. The revised gloss illustrates the factual distinction a translator needs to preserve; it need not be the best grammatical surface form in any particular target language.

Counterexample probe: Could surprise account for the letter falling? Model a plausible world in which the source remains true but the flawed gloss would be false. The resulting contrast shows exactly which agent, patient, recipient, experiencer, source of evidence, action or temporal link the poor rendering changed. This is a stronger test than simply asking whether the sentence sounds natural.

Diagnostic 57: Literature and dialogue — source-to-target event integrity

Source: “He watched the door close and wondered who had moved it.” Misleading gloss: “He closed the door and wondered who else had touched it.” Meaning-preserving gloss: “He saw the door close but did not know who moved it.” The focal character is an observer and questioner, not a confirmed actor; uncertainty drives the narrative. The revised gloss illustrates the factual distinction a translator needs to preserve; it need not be the best grammatical surface form in any particular target language.

Counterexample probe: Does the target preserve the mystery that the original creates? Model a plausible world in which the source remains true but the flawed gloss would be false. The resulting contrast shows exactly which agent, patient, recipient, experiencer, source of evidence, action or temporal link the poor rendering changed. This is a stronger test than simply asking whether the sentence sounds natural.

Diagnostic 58: Literature and dialogue — source-to-target event integrity

Source: “The song reminded her of her sister.” Misleading gloss: “She sang a song to her sister.” Meaning-preserving gloss: “Hearing or recalling the song brought her sister to mind.” The song is a trigger of memory; the sister is the mental referent, not a listener in a communication event. The revised gloss illustrates the factual distinction a translator needs to preserve; it need not be the best grammatical surface form in any particular target language.

Counterexample probe: Must the sister be present for the source to hold? Model a plausible world in which the source remains true but the flawed gloss would be false. The resulting contrast shows exactly which agent, patient, recipient, experiencer, source of evidence, action or temporal link the poor rendering changed. This is a stronger test than simply asking whether the sentence sounds natural.

Diagnostic 59: Literature and dialogue — source-to-target event integrity

Source: “The crowd carried the singer through the doorway.” Misleading gloss: “The singer carried the crowd through the doorway.” Meaning-preserving gloss: “Members of the crowd physically carried the singer inside.” Collective agent and affected person are reversed; even a dramatic literary phrase has participants. The revised gloss illustrates the factual distinction a translator needs to preserve; it need not be the best grammatical surface form in any particular target language.

Counterexample probe: Whose body crosses the threshold because of the others’ lifting? Model a plausible world in which the source remains true but the flawed gloss would be false. The resulting contrast shows exactly which agent, patient, recipient, experiencer, source of evidence, action or temporal link the poor rendering changed. This is a stronger test than simply asking whether the sentence sounds natural.

Diagnostic 60: Literature and dialogue — source-to-target event integrity

Source: “He was spoken of with admiration by people he had never met.” Misleading gloss: “He spoke admiringly about people he had never met.” Meaning-preserving gloss: “Unknown admirers praised him in conversation, although he had never met them.” The source makes him the topic of other people’s speech, not the speaker. The revised gloss illustrates the factual distinction a translator needs to preserve; it need not be the best grammatical surface form in any particular target language.

Counterexample probe: Who utters the admiration and who receives it as reputation? Model a plausible world in which the source remains true but the flawed gloss would be false. The resulting contrast shows exactly which agent, patient, recipient, experiencer, source of evidence, action or temporal link the poor rendering changed. This is a stronger test than simply asking whether the sentence sounds natural.

Diagnostic 61: Literature and dialogue — source-to-target event integrity

Source: “Mira’s apology made Elias uneasy.” Misleading gloss: “Elias apologized uneasily to Mira.” Meaning-preserving gloss: “Mira apologized; her apology caused Elias to feel uneasy.” The apology’s speaker and the emotion’s experiencer are different people. The revised gloss illustrates the factual distinction a translator needs to preserve; it need not be the best grammatical surface form in any particular target language.

Counterexample probe: Would Elias still feel uneasy if he never spoke? Model a plausible world in which the source remains true but the flawed gloss would be false. The resulting contrast shows exactly which agent, patient, recipient, experiencer, source of evidence, action or temporal link the poor rendering changed. This is a stronger test than simply asking whether the sentence sounds natural.

Diagnostic 62: Literature and dialogue — source-to-target event integrity

Source: “The guard was told to remain, yet he stepped aside.” Misleading gloss: “The guard told someone to remain, then stepped aside.” Meaning-preserving gloss: “Someone instructed the guard to stay, but the guard moved aside.” Recipient of a directive becomes sender in the flawed gloss; the contradiction between order and response disappears. The revised gloss illustrates the factual distinction a translator needs to preserve; it need not be the best grammatical surface form in any particular target language.

Counterexample probe: Does the target keep the guard’s noncompliance? Model a plausible world in which the source remains true but the flawed gloss would be false. The resulting contrast shows exactly which agent, patient, recipient, experiencer, source of evidence, action or temporal link the poor rendering changed. This is a stronger test than simply asking whether the sentence sounds natural.

Diagnostic 63: Literature and dialogue — source-to-target event integrity

Source: “Nobody knew whether the mirror had broken by itself.” Misleading gloss: “Someone broke the mirror while nobody watched.” Meaning-preserving gloss: “No one knew whether the mirror had broken on its own.” The plot deliberately withholds both cause and intentionality, rather than providing an unseen human culprit. The revised gloss illustrates the factual distinction a translator needs to preserve; it need not be the best grammatical surface form in any particular target language.

Counterexample probe: Could spontaneous fracture still be a possibility? Model a plausible world in which the source remains true but the flawed gloss would be false. The resulting contrast shows exactly which agent, patient, recipient, experiencer, source of evidence, action or temporal link the poor rendering changed. This is a stronger test than simply asking whether the sentence sounds natural.

Diagnostic 64: Digital systems and workplace — source-to-target event integrity

Source: “The repository bot merged the change after the reviewer approved it.” Misleading gloss: “The reviewer merged the change.” Meaning-preserving gloss: “A bot completed the merge after human approval.” Approval and execution are distinct actions; an audit trail needs both. The revised gloss illustrates the factual distinction a translator needs to preserve; it need not be the best grammatical surface form in any particular target language.

Counterexample probe: Could the reviewer approve from a phone while automation performs the merge later? Model a plausible world in which the source remains true but the flawed gloss would be false. The resulting contrast shows exactly which agent, patient, recipient, experiencer, source of evidence, action or temporal link the poor rendering changed. This is a stronger test than simply asking whether the sentence sounds natural.

Diagnostic 65: Digital systems and workplace — source-to-target event integrity

Source: “The project lead assigned the review to an engineer, but it was not completed.” Misleading gloss: “The project lead reviewed the code.” Meaning-preserving gloss: “The lead assigned a review that remained unfinished.” Assignment is not performance, and the named reviewer is not the assigner. The revised gloss illustrates the factual distinction a translator needs to preserve; it need not be the best grammatical surface form in any particular target language.

Counterexample probe: Can a review remain pending while the assignment record is valid? Model a plausible world in which the source remains true but the flawed gloss would be false. The resulting contrast shows exactly which agent, patient, recipient, experiencer, source of evidence, action or temporal link the poor rendering changed. This is a stronger test than simply asking whether the sentence sounds natural.

Diagnostic 66: Digital systems and workplace — source-to-target event integrity

Source: “The security service blocked the login, and the user contacted support.” Misleading gloss: “Support blocked the user’s login.” Meaning-preserving gloss: “The security service rejected the login; the user then contacted support.” An automatic control and a support interaction are separate steps. The revised gloss illustrates the factual distinction a translator needs to preserve; it need not be the best grammatical surface form in any particular target language.

Counterexample probe: Who applied the access rule, and who received the later contact? Model a plausible world in which the source remains true but the flawed gloss would be false. The resulting contrast shows exactly which agent, patient, recipient, experiencer, source of evidence, action or temporal link the poor rendering changed. This is a stronger test than simply asking whether the sentence sounds natural.

Diagnostic 67: Digital systems and workplace — source-to-target event integrity

Source: “The document was moved after approval, but the audit log did not identify who moved it.” Misleading gloss: “The approver moved the document.” Meaning-preserving gloss: “The approved document was moved, and the mover’s identity was absent from the log.” Temporal sequence and shared objects do not imply shared agents. The revised gloss illustrates the factual distinction a translator needs to preserve; it need not be the best grammatical surface form in any particular target language.

Counterexample probe: Could someone other than the approver perform the move? Model a plausible world in which the source remains true but the flawed gloss would be false. The resulting contrast shows exactly which agent, patient, recipient, experiencer, source of evidence, action or temporal link the poor rendering changed. This is a stronger test than simply asking whether the sentence sounds natural.

Diagnostic 68: Digital systems and workplace — source-to-target event integrity

Source: “The employee was invited to enroll in training.” Misleading gloss: “The employee enrolled in training.” Meaning-preserving gloss: “The employee received an invitation to enroll.” Invitation is a communicative opportunity, not completed registration. The revised gloss illustrates the factual distinction a translator needs to preserve; it need not be the best grammatical surface form in any particular target language.

Counterexample probe: Could the employee ignore the invitation? Model a plausible world in which the source remains true but the flawed gloss would be false. The resulting contrast shows exactly which agent, patient, recipient, experiencer, source of evidence, action or temporal link the poor rendering changed. This is a stronger test than simply asking whether the sentence sounds natural.

Diagnostic 69: Digital systems and workplace — source-to-target event integrity

Source: “The translation system suggested a term; a human reviewer rejected it.” Misleading gloss: “The system rejected the suggested term.” Meaning-preserving gloss: “The machine proposed a term, which a human reviewer rejected.” Suggestion and acceptance decision belong to different actors; reversing them misstates governance of AI output. The revised gloss illustrates the factual distinction a translator needs to preserve; it need not be the best grammatical surface form in any particular target language.

Counterexample probe: What should the audit record attribute to a human and what to software? Model a plausible world in which the source remains true but the flawed gloss would be false. The resulting contrast shows exactly which agent, patient, recipient, experiencer, source of evidence, action or temporal link the poor rendering changed. This is a stronger test than simply asking whether the sentence sounds natural.

Diagnostic 70: Digital systems and workplace — source-to-target event integrity

Source: “The assistant copied the meeting notes at the coordinator’s request.” Misleading gloss: “The coordinator copied the notes.” Meaning-preserving gloss: “The coordinator requested copying; the assistant performed it.” Delegation involves requester, performer and moved content; the target should retain this division. The revised gloss illustrates the factual distinction a translator needs to preserve; it need not be the best grammatical surface form in any particular target language.

Counterexample probe: Who actually made the copy, and who merely initiated the task? Model a plausible world in which the source remains true but the flawed gloss would be false. The resulting contrast shows exactly which agent, patient, recipient, experiencer, source of evidence, action or temporal link the poor rendering changed. This is a stronger test than simply asking whether the sentence sounds natural.

Diagnostic 71: Digital systems and workplace — source-to-target event integrity

Source: “The dashboard displayed a forecast that the analyst later revised.” Misleading gloss: “The dashboard revised the analyst’s forecast.” Meaning-preserving gloss: “The dashboard showed a forecast; the analyst subsequently revised it.” Display is a presentation event; revision is an independent human editorial act. The revised gloss illustrates the factual distinction a translator needs to preserve; it need not be the best grammatical surface form in any particular target language.

Counterexample probe: Which participant changes the forecast’s content? Model a plausible world in which the source remains true but the flawed gloss would be false. The resulting contrast shows exactly which agent, patient, recipient, experiencer, source of evidence, action or temporal link the poor rendering changed. This is a stronger test than simply asking whether the sentence sounds natural.

Diagnostic 72: Digital systems and workplace — source-to-target event integrity

Source: “The alert was generated when the database reached capacity, but no message reached the duty officer.” Misleading gloss: “The duty officer generated a capacity alert.” Meaning-preserving gloss: “The database triggered an alert, but the duty officer did not receive it.” Trigger, system generation, transport and human receipt must remain separate to diagnose a notification failure. The revised gloss illustrates the factual distinction a translator needs to preserve; it need not be the best grammatical surface form in any particular target language.

Counterexample probe: Could the alert exist in a log even though nobody received a message? Model a plausible world in which the source remains true but the flawed gloss would be false. The resulting contrast shows exactly which agent, patient, recipient, experiencer, source of evidence, action or temporal link the poor rendering changed. This is a stronger test than simply asking whether the sentence sounds natural.

Transfer workshops: from isolated roles to reliable multilingual review

These twenty workshops train a translator to produce a source role ledger, a natural target-language draft, a deliberately wrong rival rendering and an acceptance explanation. A learner should not answer merely by copying the English safe glosses. Choose an actual language pair, obtain suitable professional feedback for high-stakes uses, and record the range of natural target constructions. Differences in voice, case, constituent order and null arguments are expected; changes in event truth conditions are not.

Workshop 1: Active sentence to natural passive

Starting text: “The quality team rejected three faulty labels.” Task: Write a natural target version that foregrounds labels while retaining the quality team as rejecting agent. Compare with the version where labels are said merely to have been rejected; explain what institutional attribution disappears.

Acceptance test: Pass only if affected items, count, rejecting authority and event result are consistent. Produce a role ledger, at least one plausible counterexample exposing the bad alternative, and a short note explaining why your chosen target wording is idiomatic rather than a word-for-word imitation. Ask a second reader to identify the event participants without looking at the source. Where their answers differ from your ledger, revise the sentence or escalate the source ambiguity before release.

Workshop 2: Unnamed agent with deliberate uncertainty

Starting text: “The archive box was moved overnight.” Task: Translate without naming a mover or asserting an accidental movement. Contrast a truly unspecified passive with a target clause that implies staff moved it; annotate any unavoidable target-language ambiguity.

Acceptance test: The target may leave identity unknown but must not imply that no one could have moved it. Produce a role ledger, at least one plausible counterexample exposing the bad alternative, and a short note explaining why your chosen target wording is idiomatic rather than a word-for-word imitation. Ask a second reader to identify the event participants without looking at the source. Where their answers differ from your ledger, revise the sentence or escalate the source ambiguity before release.

Workshop 3: Agent versus non-agent cause

Starting text: “Heat expanded the metal collar.” Task: Preserve heat as physical cause while avoiding invented technician intervention. Test a target in which an engineer heats the collar intentionally; identify the extra action, person and control inference.

Acceptance test: Keep causal source and observed change without manufacturing a new action. Produce a role ledger, at least one plausible counterexample exposing the bad alternative, and a short note explaining why your chosen target wording is idiomatic rather than a word-for-word imitation. Ask a second reader to identify the event participants without looking at the source. Where their answers differ from your ledger, revise the sentence or escalate the source ambiguity before release.

Workshop 4: The same person in two roles

Starting text: “Eli cut himself while unpacking equipment.” Task: Track Eli as actor and affected person in one event. Compare reflexive and non-reflexive translations and determine whether target morphology requires explicit repetition.

Acceptance test: Reject any version in which a different worker is injured. Produce a role ledger, at least one plausible counterexample exposing the bad alternative, and a short note explaining why your chosen target wording is idiomatic rather than a word-for-word imitation. Ask a second reader to identify the event participants without looking at the source. Where their answers differ from your ledger, revise the sentence or escalate the source ambiguity before release.

Workshop 5: Multiple actors in one sentence

Starting text: “Nia drew the plan; Ezra checked it; Sam approved it.” Task: Annotate author, reviewer and approver as three event roles. Resist the tempting compressed form saying the team drew and approved the plan; identify whether legal or organizational attribution would change.

Acceptance test: Three predicates must keep their named responsible participants. Produce a role ledger, at least one plausible counterexample exposing the bad alternative, and a short note explaining why your chosen target wording is idiomatic rather than a word-for-word imitation. Ask a second reader to identify the event participants without looking at the source. Where their answers differ from your ledger, revise the sentence or escalate the source ambiguity before release.

Workshop 6: Role of a machine

Starting text: “The scanner flagged the barcode for manual review.” Task: Keep a machine detection stage separate from later human assessment. Test whether the target asserts review already happened or identifies a reviewer who has not acted.

Acceptance test: Preserve automatic flagging, future review and unknown individual reviewer. Produce a role ledger, at least one plausible counterexample exposing the bad alternative, and a short note explaining why your chosen target wording is idiomatic rather than a word-for-word imitation. Ask a second reader to identify the event participants without looking at the source. Where their answers differ from your ledger, revise the sentence or escalate the source ambiguity before release.

Workshop 7: Recipient and beneficiary

Starting text: “A grant was paid to the library for its language program.” Task: Separate payer, recipient, purpose and possible beneficiaries. Check whether a translation implies that the library granted money or that a specific person directly received funds.

Acceptance test: Maintain direction of transfer and stated purpose without inventing a donor. Produce a role ledger, at least one plausible counterexample exposing the bad alternative, and a short note explaining why your chosen target wording is idiomatic rather than a word-for-word imitation. Ask a second reader to identify the event participants without looking at the source. Where their answers differ from your ledger, revise the sentence or escalate the source ambiguity before release.

Workshop 8: Verbs of perception

Starting text: “Uma heard a latch click before seeing the door move.” Task: Annotate auditory experiencer, sound source, visual experiencer and the two events. Test translation alternatives that turn hearing into active listening or clicking into deliberate unlatching.

Acceptance test: Preserve the perceptual timeline and avoid assuming who touched the latch. Produce a role ledger, at least one plausible counterexample exposing the bad alternative, and a short note explaining why your chosen target wording is idiomatic rather than a word-for-word imitation. Ask a second reader to identify the event participants without looking at the source. Where their answers differ from your ledger, revise the sentence or escalate the source ambiguity before release.

Workshop 9: Causative versus commissioned action

Starting text: “The owner had the mural cleaned by specialists.” Task: Distinguish arranger, actual cleaners and affected artwork. Compare one target with a lexical causative and another with two clauses; consider what each says about physical performance.

Acceptance test: Do not make the owner personally scrub the mural. Produce a role ledger, at least one plausible counterexample exposing the bad alternative, and a short note explaining why your chosen target wording is idiomatic rather than a word-for-word imitation. Ask a second reader to identify the event participants without looking at the source. Where their answers differ from your ledger, revise the sentence or escalate the source ambiguity before release.

Workshop 10: Attempt versus accomplishment

Starting text: “The crew tried to open the hatch but failed.” Task: Express intention, attempt and unsuccessful result. Check that a target connective does not imply a successful partial opening or final release.

Acceptance test: The hatch must be allowed to remain closed in a consistent possible world. Produce a role ledger, at least one plausible counterexample exposing the bad alternative, and a short note explaining why your chosen target wording is idiomatic rather than a word-for-word imitation. Ask a second reader to identify the event participants without looking at the source. Where their answers differ from your ledger, revise the sentence or escalate the source ambiguity before release.

Workshop 11: Speech reporting and epistemic limits

Starting text: “The witness said the guard blocked the exit.” Task: Retain the witness as speaker and the guard as agent inside a reported proposition. Produce a target version compatible with the witness being mistaken.

Acceptance test: Do not turn reported speech into the narrator’s independently verified claim. Produce a role ledger, at least one plausible counterexample exposing the bad alternative, and a short note explaining why your chosen target wording is idiomatic rather than a word-for-word imitation. Ask a second reader to identify the event participants without looking at the source. Where their answers differ from your ledger, revise the sentence or escalate the source ambiguity before release.

Workshop 12: Accountability in governance

Starting text: “The deputy signed for the director, who had authorized the document.” Task: Keep signature, representation and authorization as different actions. Ask whether the deputy personally authorized the content or whether the director physically signed it.

Acceptance test: The target must preserve the actual signatory and actual authorizer. Produce a role ledger, at least one plausible counterexample exposing the bad alternative, and a short note explaining why your chosen target wording is idiomatic rather than a word-for-word imitation. Ask a second reader to identify the event participants without looking at the source. Where their answers differ from your ledger, revise the sentence or escalate the source ambiguity before release.

Workshop 13: Collaboration and shared events

Starting text: “The three workers moved a cabinet together.” Task: Track the collective action without overasserting the exact distribution of strength. Explore target-language group markers, conjunction and plural verbs while checking whether one person could have only guided.

Acceptance test: Preserve joint participation without inventing a precise mechanical breakdown. Produce a role ledger, at least one plausible counterexample exposing the bad alternative, and a short note explaining why your chosen target wording is idiomatic rather than a word-for-word imitation. Ask a second reader to identify the event participants without looking at the source. Where their answers differ from your ledger, revise the sentence or escalate the source ambiguity before release.

Workshop 14: Intent and accident

Starting text: “The child knocked the vase over while turning around.” Task: Avoid adding willful destruction or strict exculpation not supported by the text. Compare an accident-neutral rendering with a target verb that implies intentional striking.

Acceptance test: Keep the movement, contact and resulting fall; leave intention at its source level. Produce a role ledger, at least one plausible counterexample exposing the bad alternative, and a short note explaining why your chosen target wording is idiomatic rather than a word-for-word imitation. Ask a second reader to identify the event participants without looking at the source. Where their answers differ from your ledger, revise the sentence or escalate the source ambiguity before release.

Workshop 15: Process result and responsible party

Starting text: “The file became unreadable after conversion.” Task: Preserve a damaged result and temporal relation without naming an operator. Check whether conversion necessarily caused the problem or merely preceded its discovery.

Acceptance test: The target must not falsely establish malicious alteration or a known person responsible. Produce a role ledger, at least one plausible counterexample exposing the bad alternative, and a short note explaining why your chosen target wording is idiomatic rather than a word-for-word imitation. Ask a second reader to identify the event participants without looking at the source. Where their answers differ from your ledger, revise the sentence or escalate the source ambiguity before release.

Workshop 16: Instructions and delegated action

Starting text: “The supervisor instructed the assistant to print the receipt.” Task: Keep communicator, addressee and proposed performance distinct. Test whether a target perfective verb unintentionally asserts that printing occurred.

Acceptance test: If there is no following confirmation, represent the act as an instruction rather than a completed print job. Produce a role ledger, at least one plausible counterexample exposing the bad alternative, and a short note explaining why your chosen target wording is idiomatic rather than a word-for-word imitation. Ask a second reader to identify the event participants without looking at the source. Where their answers differ from your ledger, revise the sentence or escalate the source ambiguity before release.

Workshop 17: Participant roles in negation

Starting text: “The committee did not authorize the purchase.” Task: Preserve the negation of committee authorization without concluding that somebody else approved. Test whether the target permits a world in which nobody authorized the purchase.

Acceptance test: Reject a translation that reverses authorization or invents a substitute decision-maker. Produce a role ledger, at least one plausible counterexample exposing the bad alternative, and a short note explaining why your chosen target wording is idiomatic rather than a word-for-word imitation. Ask a second reader to identify the event participants without looking at the source. Where their answers differ from your ledger, revise the sentence or escalate the source ambiguity before release.

Workshop 18: Conditional chain of events

Starting text: “If the sensor detects smoke, the controller will close the intake.” Task: Identify sensor as detector, controller as actuator and closing as conditional future action. Compare a target where the sensor closes the intake itself, or closure has already occurred.

Acceptance test: Preserve condition, future modality and division of system functions. Produce a role ledger, at least one plausible counterexample exposing the bad alternative, and a short note explaining why your chosen target wording is idiomatic rather than a word-for-word imitation. Ask a second reader to identify the event participants without looking at the source. Where their answers differ from your ledger, revise the sentence or escalate the source ambiguity before release.

Workshop 19: Unknown versus intentionally omitted agent

Starting text: “The painting was removed from the exhibition.” Task: Write two contextual notes: one where the agent is truly unknown and one where the institution withholds the identity. Decide whether a target impersonal construction preserves both possibilities or strongly implies one.

Acceptance test: Translation should report removal accurately and flag context-specific ambiguity instead of guessing. Produce a role ledger, at least one plausible counterexample exposing the bad alternative, and a short note explaining why your chosen target wording is idiomatic rather than a word-for-word imitation. Ask a second reader to identify the event participants without looking at the source. Where their answers differ from your ledger, revise the sentence or escalate the source ambiguity before release.

Workshop 20: Cross-paragraph continuity

Starting text: “At dawn Asha alerted Benoit. Later he called the operator. She acknowledged the fault.” Task: Track pronouns by testing referents against preceding discourse. Prepare alternatives where “she” could refer to Asha or a newly introduced female operator; annotate what the text supports.

Acceptance test: Do not solve a referential ambiguity by altering the characters’ actual actions. Produce a role ledger, at least one plausible counterexample exposing the bad alternative, and a short note explaining why your chosen target wording is idiomatic rather than a word-for-word imitation. Ask a second reader to identify the event participants without looking at the source. Where their answers differ from your ledger, revise the sentence or escalate the source ambiguity before release.

Four-week implementation plan for a real translation team

Week 1: establish a baseline and a shared role vocabulary

Choose approximately thirty short sentences from source materials already cleared for internal review. Include event statements, requests, physical-process descriptions, customer communications and reports of speech. Remove private information and keep the authentic structural challenges. Have two reviewers independently annotate action, agent, patient, experiencer, recipient, instrument, cause, intent, known/unknown roles and resulting state. Compare disagreement before discussing wording. The aim is not to enforce universal labels where semantic theories differ; it is to discover which distinctions influence practical meaning. Create a one-page local annotation convention and a worked reference set.

Week 2: translate and construct rival readings

Translate the baseline into the actual target language, then ask reviewers to produce one plausible but incorrect rival meaning for each high-risk example. A useful rival often changes a single participant or event stage: author becomes messenger, request becomes fulfillment, witness becomes perpetrator, system becomes operator, or perception becomes intentional action. Reviewers should write a counterexample world proving the mismatch. Maintain a spreadsheet with source identifier, event-role ledger, actual target rendering, proposed correction, context evidence and owner of the decision. Do not accept a translation as tested merely because it passed spelling checks.

Week 3: integrate into an editorial workflow

Add role checks to the project’s existing proofreading or review stage. Risk-rank the cases: messages about safety, consent, financial responsibility, official attribution or operational control need closer inspection than ordinary stylistic prose. Ask a native target-language reader a short comprehension question before showing the source. Their reconstruction of who acted and what happened is valuable evidence. If a translation fails, fix the narrow role relation rather than rewriting unrelated paragraphs. For an unresolved source ambiguity, file an author query; for an intentionally unresolved narrative ambiguity, preserve it.

Week 4: retest, sample and make the system sustainable

Use a new thirty-sentence set that differs in topic and grammar from the baseline. Track critical and major role errors separately from stylistic suggestions, so the team does not claim quality improvement merely by increasing review volume. Ask whether corrections survive revisions and whether repeated errors cluster around one language pair, one template or one machine-translation setting. Produce a short release note explaining the chosen review scope, any unresolved issues and the evidence supporting sign-off. The team should then integrate the compact role ledger into onboarding, translation briefs and subsequent audit samples.

Operational quality rubric and release gates

Critical: A target invents or hides an actor in a way likely to alter safety decisions, consent, rights, liability, authorization or necessary action. Examples include making a user the performer of a safety procedure assigned to a professional, or turning “not yet approved” into “approved.” Stop release and involve the responsible domain specialist.

Major: The target materially reverses an agent–patient relation, loses an essential recipient, changes a reported allegation into a fact, transforms a request into completion, or mistakes an instrument for a decision-maker. Correct and review surrounding occurrences of the same structure.

Moderate: The target preserves basic events but weakens recoverable attribution, overstates a nuance of volition, or forces an unnatural form likely to confuse readers. Correct where the effect is material for the intended task.

Minor or stylistic: Word order, preference for active or passive, and degree of explicitness vary naturally without affecting roles, truth conditions or meaningful discourse focus. Record a style preference only after proving that no participant relationship changed.

The labels above are an editorial proposal rather than an official scoring schedule. ISO 5060:2024 provides guidance on human evaluation of translation output, including analytic approaches with error types, penalty points and evaluation settings; the document should not be portrayed as prescribing the custom ledger or severity thresholds developed here. Likewise, MQM gives a flexible taxonomy for accuracy assessment, not an automatic verdict on a particular sentence.

Glossary of participant roles, grammar and translation diagnostics

Agent: An event participant commonly construed as intentionally initiating an action; not synonymous with grammatical subject.
Patient: An event participant affected by an action or undergoing change.
Theme: The participant moved, located, possessed or otherwise involved, often without requiring a change of state.
Experiencer: A participant who senses, feels, knows, remembers or otherwise undergoes a mental experience.
Stimulus: What prompts or is perceived in a mental or sensory event.
Instrument: A tool or means used to bring about an event; its appearance as subject does not prove independent will.
Causer: A participant or condition causally responsible for a change, whether intentional or not.
Recipient: The endpoint of a transfer, message or provision.
Beneficiary: Someone who gains advantage from an event, not necessarily the one who performs it.
Source: The origin of a transfer, information or motion, interpreted according to the event frame.
Goal: The intended or achieved destination of motion or other directed activity.
Grammatical subject: A syntactic position or relation; can be agent, patient, theme or experiencer.
Voice: A grammatical system affecting how participant roles are represented and foregrounded.
Active voice: A clause pattern often, but not invariably, placing an agent-like participant in subject position.
Passive voice: A clause pattern that often foregrounds an affected entity and backgrounds or demotes the agent.
Valency: The argument-structure possibilities associated with a predicate.
Argument: A participant expression licensed or expected by a predicate’s meaning and syntax.
Adjunct: A modifier adding context such as time, location or manner without functioning as a core argument.
Causative: A construction representing one participant or event as causing another event or state.
Anticausative: A construction describing a change of state without necessarily representing an external causer.
Impersonal construction: A clause form whose surface syntax does not identify an ordinary referential subject.
Reflexive: A relation where a participant is linked back to another role of the same entity within the relevant construction.
Reciprocal: A relation involving participants acting toward one another.
Collective action: An event attributed to a group acting as a unit, not necessarily with identical contribution from each member.
Distributive action: An event or property construed as applying individually across members.
Coreference: Different expressions in discourse that pick out the same entity.
Null argument: A participant not overtly pronounced or written where the language or context permits its omission.
Agent suppression: Omission or backgrounding of a performer in a way the grammar and discourse allow.
Metonymic institutional agent: An organization named as actor although individuals act through its procedures.
Control: A relation determining who is understood as the subject of an embedded or infinitival event.
Raising: A syntactic pattern in which an argument appears with a predicate that does not assign it an ordinary agentive role.
Event frame: A structured representation of an event type and its associated participants and relations.
Event entailment: A truth-preserving implication that a given event or result occurred.
Presupposition: Information presented as backgrounded or taken for granted, not necessarily newly asserted.
Entailment test: A check whether the truth of one statement requires the truth of another.
Counterexample world: A plausible scenario where one wording is true and a supposedly equivalent wording is false.
Evidential source: The origin of a claim or perception: direct observation, report, measurement, inference or other channel.
Information structure: How a sentence organizes given, new, topical or contrastive information.
Mistranslation: A target expression that inaccurately represents source content.
Over-translation: The introduction of specificity, detail or commitments not supported by the source.
Under-translation: Loss of meaningful source distinctions in the target.
Role ledger: The practical event-by-event annotation scheme developed in this article to support review.

Frequently asked questions

1. What is the difference between an agent and a patient in translation?

An agent is typically the instigating participant of an action; a patient is the participant affected by that action. The grammatical subject need not be the agent. A translation must keep the event roles even when subject order or verbal voice changes.

2. Does passive voice always hide who performed an action?

No. Many passives allow an overt by-phrase or other agent expression. An omitted agent can also be recoverable from context. The safe question is whether source and target license the same inference about agent identity.

3. Must I translate every passive with a passive?

No. Different languages structure events differently. A natural active, impersonal or intransitive form may be preferable if it preserves agent status, affected participant, causation and discourse purpose.

4. How can I translate passive voice when the target language has no equivalent?

Represent the original event and any known or unknown agents using a construction licensed by the target language. Ask a native reviewer what the target implies about who acted. Never invent a performer merely to fill a grammatical subject slot.

5. Why does the difference between ‘opened’ and ‘was opened’ matter?

In context, an inchoative change-of-state report can leave causes open while a passive may suggest an action affecting the object. These readings are language- and construction-sensitive, so test the actual clause and context instead of relying on a single English heuristic.

6. What is a semantic role ledger?

It is a short event record specifying event, causer, agent, affected entity, experiencer, recipient, instrument, result, intent and unknowns. It helps compare the meanings expressed by source and target.

7. Can a machine be an agent?

Everyday language sometimes makes machines the subjects of active predicates. Whether that is a grammatical agent, a semantic causer or metonymic shorthand depends on the event. Do not infer autonomous intention simply from subject position.

8. Is an institution an actor?

Institutions can be named as actors in ordinary and formal texts. Preserve that level of attribution; do not automatically replace a board or organization with its chair or an unnamed employee.

9. How do I avoid blaming someone in translation?

Separate event, cause, intention, knowledge, control and responsibility. If the source confirms only a malfunction, do not add a human culprit. If the source explicitly attributes an action, do not erase the actor to avoid an uncomfortable sentence.

10. What is the difference between a recipient and a beneficiary?

A recipient receives a transferred item or message. A beneficiary benefits from an event. A person may be both, but the source must support the relevant relationship.

11. How are causatives handled across languages?

Identify who arranges, orders, permits, enables or directly carries out an action. The target may use different syntax, but should not confuse arranging a repair with personally repairing something.

12. How should I translate unspecified agents?

Choose natural target wording that maintains the appropriate level of uncertainty. If it is important whether the agent is unknown, withheld or recoverable, record that distinction and review the wider passage.

13. What should I do with ambiguous pronouns?

Check the full antecedent chain before naming anyone. Where the source truly permits more than one reading, retain that ambiguity if natural and safe, or ask the author for clarification.

14. Are grammatical dependency parsers sufficient?

No. Dependency labels may reveal subjects and passive constructions, but they cannot by themselves resolve every event role, intentionality, implied participant or factual attribution. Use them as review cues.

15. Can machine translation preserve agent–patient roles?

Often, but fluent machine output can also swap actors, omit agents or convert requests into completed actions. Independently reconstruct target meaning and compare with the source, especially in consequential content.

16. How does this method help subtitles?

Subtitle compression tends to drop repeated names or clauses. Preserve at least the identities and actions needed for a viewer to know who requested, refused, observed or performed the event.

17. Does every omission count as an error?

No. Languages legitimately omit recoverable arguments. It becomes a substantive error when omission changes the reader’s likely understanding of essential participant roles or source claims.

18. What is the difference between a reported allegation and a fact?

A report states that a speaker made a claim; it does not necessarily certify that the underlying event happened. Translation must keep the speaker, reported proposition and evidential status distinct.

19. How can I test a translation without knowing formal logic?

Invent a plausible situation where the original sentence is true. Ask whether the target would also be true. Swap the possible actor or change intention to expose unsupported assumptions.

20. Should every article use the same severity scale?

No. Establish risk criteria for each project, language pair and use case. Use a consistent audit trail, but avoid treating the illustrative severity labels in this chapter as an official external standard.

21. What if the source itself uses a misleading passive?

Translate what the source actually asserts, while escalating a source-quality issue through the relevant editorial or client process. Do not silently change allegations or responsibility under the guise of fixing grammar.

22. Where does vocabulary learning fit?

Lexical selection matters when a verb encodes a particular event frame, but this chapter is not a replacement for vocabulary instruction. A person may know every word and still swap sender and recipient across languages.

23. Where does English grammar instruction fit?

Knowing active and passive grammar helps identify candidates for review. Cross-language participant fidelity adds separate tests for roles, causal relations, implied agents and context; these specialist tests are this chapter’s scope.

24. How should a team document unresolved cases?

Record source quotation, event ledger, candidate translations, ambiguity or error type, counterexample, expert question, decision and reviewer. Reopen the entry when the source changes.

25. What is the single fastest practical check?

Ask the reader of the target text who acted, what was affected and what remains unknown. Compare their answers with the source ledger. A changed answer is a signal for further investigation.

Research and terminology references

These sources establish terminology, analytical distinctions and translation-evaluation concepts. They do not endorse the particular role-ledger system, diagnostic examples or severity thresholds introduced here, which are original editorial tools. Readers should verify the current edition and applicability of each reference when using it in a formal translation project.

International Computer Science Institute — FrameNet Project. Frame semantics models event types and the participants involved; useful background for understanding why word senses require different participant configurations.
Universal Dependencies — UD v2 relations. Official cross-linguistic grammatical-relation documentation, including subject, object, passive-subject and agent-related labels. These are grammatical annotation aids, not a complete semantic-role oracle.
World Atlas of Language Structures — Feature 107A: Passive Constructions. Language-typological guidance emphasizing that English-like passive structures are not an automatic universal template.
MQM Council — Multidimensional Quality Metrics Full Typology. A framework for categorizing translation-quality problems, especially accuracy, addition, omission and mistranslation.
ISO 5060:2024 — Translation services: Evaluation of translation output. The ISO description covers guidance for evaluation of human, post-edited and unedited machine translation output, with analytic error-based approaches and evaluator competence.
ISO 17100:2015 — Translation services requirements. High-level process and resource context; not a source for the article’s invented examples or a claim that the present chapter ensures certification.
W3C — Internationalization Tag Set 2.0. Useful for translation and localization metadata, including which content is or is not intended to be translated, supporting text processing across language boundaries.

Next routes: translation architecture, vocabulary and English structure

This is chapter EDKSG-TRANS-MASTER-WORLD-1430 of the specialist meaning-preservation architecture. For the broader system—source understanding, context, terminology, revisions, translation methods and quality control—return to Master Art of Translation | The Complete System for Moving Meaning Between Languages. For the immediately preceding specialist problem of baselines, denominators and comparative reference groups, see Master Art of Translation | The Comparative Frame Fidelity System. These are related but have different editorial ownership.

When the main obstacle is understanding the sense of an unfamiliar verb or building lexical precision, use the protected Vocabulary Learning Hub as a separate learning resource. When the obstacle is how English grammar expresses active or passive constructions, use How English Works | The Complete Map and Learning Routes or its existing voice-and-agency explanation. These owners are not reproduced or replaced here. This chapter’s original work concerns comparing participant mappings between source and target languages and testing the consequences of changing those mappings.

Final release rule: first reconstruct the event, then preserve the participants, then write naturally, then challenge the translation with counterexamples. A target sentence can be shorter, longer, active, passive or impersonal; it cannot fairly change who is known to have acted, who was affected, which process happened, or whether the source deliberately leaves those facts unresolved.