VIEW THIS AS

Auto mode follows the Route Engine until you choose a viewpoint.

YOU ARE HERE

ROUTE CHECK

CONNECTED TO

WHAT NEXT

Use the canonical route for this room, or HELP if you are unsure.

Master Art of Translation | The Temporal Meaning System — How Tense, Aspect, Event Order, Deadlines and Time Zones Preserve When Things Happen

How to translate tense, aspect and temporal reference accurately

How do you translate tense and aspect without losing the time of an event? Translators need to preserve the sequence of actions, the difference between ongoing and completed work, a speaker’s reference point, and whether a date is a forecast, a deadline or a recorded fact. This comprehensive guide explains temporal meaning in translation with practical methods for translating past-perfect relationships, aspect, reported speech, relative dates, event order, duration, deadlines, date formatting and time zones.

The central problem is that a fluent translation can move an event. “The test had finished when the alarm sounded” does not assert the same chronology as “the test finished after the alarm.” “The team has been reviewing the results” does not certify that the review is complete. “Before Friday,” “by Friday” and “until Friday” may carry different boundary rules, while “tomorrow” depends on who said it and when. An apparently minor change may cause a reader to miss a deadline, rely on an unverified result or misread a historical sequence.

This article presents a repeatable temporal translation quality assurance system for readers, editors, translators and multilingual product teams. It begins with search-intent answers, diagnoses the hidden problem, explains the linguistic and technical mechanisms, establishes a structured method, examines failure modes and develops original worked cases. Its methods are world-facing: examples are explained in English, but no target language is presumed to reproduce English tense forms. The final standard of accuracy is what a qualified target-language reader understands, not the number of matching words.

Immediate answer: translate the relationship, not the verb label

For every passage containing a time-related claim, ask six questions before drafting: What happened? At what reference point is it being described? What happened before, during or after it? Is it complete, ongoing, habitual, planned or uncertain? Are calendar fields, durations and boundaries exact? What decision does the statement authorize? Record the answers in a temporal contract. Then choose the natural linguistic structure that communicates those answers in the target language. The method respects languages with distinct aspect systems, optional tense marking and context-dependent event sequencing.

A minimal event graph might read: SOURCE EVENT = inspection; REFERENCE = arrival of the warning; RELATION = inspection BEFORE warning; STATUS = completed; EVIDENCE = source sentence; EXACT TIME = not supplied. A target can use a time adverb or different clause order and still be correct. It fails if the target permits the ordinary inference that the inspection began only once the warning appeared. Recording that contract makes later revisions and independent reviews possible.

The hidden problem: time travels through the entire document

A source sentence can rely on an earlier event as its temporal reference. A diary can quote tomorrow from decades earlier. A machine-generated interface can display a valid instant using an ambiguous local date string. A numbered procedure can imply sequencing even without a visible before or after. A translation memory can recycle an otherwise accurate sentence into a context with a different speaker date. Thus temporal errors are not confined to tense: they may arise from document context, version history, data fields, scheduling rules, or the pragmatics of what a reader believes has already happened.

Temporal information has at least two layers. The linguistic layer carries event identity, state, sequence, certainty and reference time. The technical layer carries timestamps, local civil-time rules, calendar settings, display patterns and recurrence definitions. Neither layer is a substitute for the other. A correct conversion from a stored UTC instant cannot decide whether “early next week” was meant as a firm commitment. A perfect translation of that phrase cannot make a missing time-zone identifier appear. The reviewer must know which kind of problem is being solved.

Mechanism: separate speech time, reference time, event time and knowledge time

Speech time is when an utterance occurs; event time is when a situation happens or holds; reference time is the vantage point from which the event is viewed. Knowledge time is when evidence became available to a speaker or institution. In a narrative, the order of telling may differ from the order of events. These distinctions help prevent historical forecasts from becoming present promises, and later discoveries from being credited to earlier decision-makers.

For example, “By the meeting, the equipment had cooled” identifies a meeting as a reference point and locates completed cooling earlier. “During the meeting, the equipment cooled” instead places an event within the meeting interval. The target language may mark these relations by different grammar or by a time phrase. The relevant cross-language invariant is the before–after or overlap relation; it is not the English auxiliary had.

Tense typically locates a situation relative to a temporal vantage point; aspect concerns its internal temporal contour, such as ongoing activity, habitual recurrence and completion. Languages do not necessarily share a one-to-one inventory of these categories. A speaker can report a planned action, an actual event, an unfinished process, an achieved result or a prolonged condition with various constructions. Translators should test those readings explicitly and avoid importing an outcome that the source never asserted.

Intervals require additional care. An instant is different from a duration. A calendar date is not always one fixed instant for readers in different zones. A recurrence such as every working day depends on a calendar and exceptions. A deadline may have an inclusive end or an exclusive cutoff. “Roughly two hours” is not an exact number of minutes to the second. A date-only string often needs an authoritative rule before a system can calculate a worldwide timestamp.

The official ISO 24617-1:2012 SemAF-Time / ISO-TimeML standard supplies a formal semantic annotation framework for time and events, confirmed as current in 2023. The practical method here simplifies such analysis into event nodes, reference points and relations like BEFORE, AFTER, OVERLAPS, CONTAINS, STARTS, ENDS and UNKNOWN. An unresolved relationship should remain unknown. It should not be converted into a confident chronological assertion merely because the translator expects events to follow a familiar order.

IETF RFC 3339 defines a profile for Internet timestamps, while RFC 9557, published April 2024, extends timestamps with additional information such as time-zone annotations and updates the treatment of Z in a specified way. A numeric UTC offset is not a permanent civil-time-zone rule. For localized display, Unicode CLDR, UTS #35 Part 4 describes date and time patterns, intervals and relative-date joining forms. These standards guide representation and formatting, but none can invent an author’s missing intention.

The repeatable sixteen-gate workflow

Gate 1: Identify the reader action

Write the action or inference the passage supports: follow a procedure, attend an event, read an incident timeline, understand a story or judge a deadline. This is an explicit verification stage rather than an abstract instruction to be careful. Record the evidence that supports the temporal interpretation and explain why the target reader should reach the same conclusion. If another reviewer cannot reproduce the decision, the passage has not been sufficiently verified. A fluent translation can remain a candidate, but a high-consequence timing assertion should not pass without a supported reference point and boundary rule.

Gate 2: Find the controlling source

Record the approved version, publication date, quote date and any schedule or logged event. Never assume the newest visible timestamp belongs to the event. This is an explicit verification stage rather than an abstract instruction to be careful. Record the evidence that supports the temporal interpretation and explain why the target reader should reach the same conclusion. If another reviewer cannot reproduce the decision, the passage has not been sufficiently verified. A fluent translation can remain a candidate, but a high-consequence timing assertion should not pass without a supported reference point and boundary rule.

Gate 3: Mark time-bearing language

Underline verb forms, conjunctions, temporal adverbs, negatives, recurrence expressions and boundaries; also inspect neighbouring sentences. This is an explicit verification stage rather than an abstract instruction to be careful. Record the evidence that supports the temporal interpretation and explain why the target reader should reach the same conclusion. If another reviewer cannot reproduce the decision, the passage has not been sufficiently verified. A fluent translation can remain a candidate, but a high-consequence timing assertion should not pass without a supported reference point and boundary rule.

Gate 4: Locate speech or quotation time

Distinguish the author, quoted speaker and time of report. Old tomorrow and now cannot automatically move to the new publication date. This is an explicit verification stage rather than an abstract instruction to be careful. Record the evidence that supports the temporal interpretation and explain why the target reader should reach the same conclusion. If another reviewer cannot reproduce the decision, the passage has not been sufficiently verified. A fluent translation can remain a candidate, but a high-consequence timing assertion should not pass without a supported reference point and boundary rule.

Gate 5: Mark event time and reference time

Identify the event, the temporal vantage point and whether a source actually supplies a precise instant. This is an explicit verification stage rather than an abstract instruction to be careful. Record the evidence that supports the temporal interpretation and explain why the target reader should reach the same conclusion. If another reviewer cannot reproduce the decision, the passage has not been sufficiently verified. A fluent translation can remain a candidate, but a high-consequence timing assertion should not pass without a supported reference point and boundary rule.

Gate 6: Create event edges

Record BEFORE, AFTER, OVERLAPS, CONTAINS or UNKNOWN for materially related events. Do not fill unsupported edges. This is an explicit verification stage rather than an abstract instruction to be careful. Record the evidence that supports the temporal interpretation and explain why the target reader should reach the same conclusion. If another reviewer cannot reproduce the decision, the passage has not been sufficiently verified. A fluent translation can remain a candidate, but a high-consequence timing assertion should not pass without a supported reference point and boundary rule.

Gate 7: Distinguish state and aspect

Label finished, in progress, habitual, merely intended, predicted, disputed, cancelled or not yet begun. This is an explicit verification stage rather than an abstract instruction to be careful. Record the evidence that supports the temporal interpretation and explain why the target reader should reach the same conclusion. If another reviewer cannot reproduce the decision, the passage has not been sufficiently verified. A fluent translation can remain a candidate, but a high-consequence timing assertion should not pass without a supported reference point and boundary rule.

Gate 8: Record certainty

Separate observation from expectation, suspicion, forecast and source-owner instruction. A predicted event is not proof of fulfillment. This is an explicit verification stage rather than an abstract instruction to be careful. Record the evidence that supports the temporal interpretation and explain why the target reader should reach the same conclusion. If another reviewer cannot reproduce the decision, the passage has not been sufficiently verified. A fluent translation can remain a candidate, but a high-consequence timing assertion should not pass without a supported reference point and boundary rule.

Gate 9: Preserve duration and units

Check elapsed hours, working days, calendar months, approximate ranges and whether measurement is at least or at most. This is an explicit verification stage rather than an abstract instruction to be careful. Record the evidence that supports the temporal interpretation and explain why the target reader should reach the same conclusion. If another reviewer cannot reproduce the decision, the passage has not been sufficiently verified. A fluent translation can remain a candidate, but a high-consequence timing assertion should not pass without a supported reference point and boundary rule.

Gate 10: Define boundary inclusivity

Clarify before, by, through, until, from and within where the precise cutoff matters. Ask the owner if the governing source is ambiguous. This is an explicit verification stage rather than an abstract instruction to be careful. Record the evidence that supports the temporal interpretation and explain why the target reader should reach the same conclusion. If another reviewer cannot reproduce the decision, the passage has not been sufficiently verified. A fluent translation can remain a candidate, but a high-consequence timing assertion should not pass without a supported reference point and boundary rule.

Gate 11: Preserve technical instant and zone

Keep structured date, UTC instant, numeric offset, named civil-time zone, calendar system and recurrence rule conceptually separate. This is an explicit verification stage rather than an abstract instruction to be careful. Record the evidence that supports the temporal interpretation and explain why the target reader should reach the same conclusion. If another reviewer cannot reproduce the decision, the passage has not been sufficiently verified. A fluent translation can remain a candidate, but a high-consequence timing assertion should not pass without a supported reference point and boundary rule.

Gate 12: Compose natural target wording

Use target-language resources, not a rigid English tense-conversion chart; preserve the underlying contract. This is an explicit verification stage rather than an abstract instruction to be careful. Record the evidence that supports the temporal interpretation and explain why the target reader should reach the same conclusion. If another reviewer cannot reproduce the decision, the passage has not been sufficiently verified. A fluent translation can remain a candidate, but a high-consequence timing assertion should not pass without a supported reference point and boundary rule.

Gate 13: Review interpolated text

Check placeholder preservation, full-sentence grammar, locale patterns, screen-reader descriptions and rendered date ranges. This is an explicit verification stage rather than an abstract instruction to be careful. Record the evidence that supports the temporal interpretation and explain why the target reader should reach the same conclusion. If another reviewer cannot reproduce the decision, the passage has not been sufficiently verified. A fluent translation can remain a candidate, but a high-consequence timing assertion should not pass without a supported reference point and boundary rule.

Gate 14: Conduct blind target-reading test

Have a proficient reader reconstruct the timeline and permitted action without reading the source or the translator’s explanation. This is an explicit verification stage rather than an abstract instruction to be careful. Record the evidence that supports the temporal interpretation and explain why the target reader should reach the same conclusion. If another reviewer cannot reproduce the decision, the passage has not been sufficiently verified. A fluent translation can remain a candidate, but a high-consequence timing assertion should not pass without a supported reference point and boundary rule.

Gate 15: Use counterfactual cases

Swap before and after, plan and result or still and no longer in a test copy; identify which distinction the target must preserve. This is an explicit verification stage rather than an abstract instruction to be careful. Record the evidence that supports the temporal interpretation and explain why the target reader should reach the same conclusion. If another reviewer cannot reproduce the decision, the passage has not been sufficiently verified. A fluent translation can remain a candidate, but a high-consequence timing assertion should not pass without a supported reference point and boundary rule.

Gate 16: Release with traceable evidence

Store source version, event graph, target text, clarifications, reader responses and review decision; repeat after relevant changes. This is an explicit verification stage rather than an abstract instruction to be careful. Record the evidence that supports the temporal interpretation and explain why the target reader should reach the same conclusion. If another reviewer cannot reproduce the decision, the passage has not been sufficiently verified. A fluent translation can remain a candidate, but a high-consequence timing assertion should not pass without a supported reference point and boundary rule.

Twenty-seven recurring failure modes

Failure 1: An earlier past event becomes a later one

A past-perfect-like relation is flattened into a sequence contradicted by the source. Review its effect on a real reader: would the person act too early, miss a valid interval, believe a process finished, or misplace an event? The repair should preserve both source-backed information and source uncertainty. Classify the failure as event order, aspect, reference anchor, recurrence, boundary, technical representation or evidence time. Compare an independent target-only interpretation with the approved source. If the difference affects safety, permissions or compliance, obtain authorized domain review before publication.

Failure 2: Work in progress becomes completed work

An activity verb is converted to a certified finished result. Review its effect on a real reader: would the person act too early, miss a valid interval, believe a process finished, or misplace an event? The repair should preserve both source-backed information and source uncertainty. Classify the failure as event order, aspect, reference anchor, recurrence, boundary, technical representation or evidence time. Compare an independent target-only interpretation with the approved source. If the difference affects safety, permissions or compliance, obtain authorized domain review before publication.

Failure 3: Habit becomes a single episode

Regular instructions are translated as one appointment. Review its effect on a real reader: would the person act too early, miss a valid interval, believe a process finished, or misplace an event? The repair should preserve both source-backed information and source uncertainty. Classify the failure as event order, aspect, reference anchor, recurrence, boundary, technical representation or evidence time. Compare an independent target-only interpretation with the approved source. If the difference affects safety, permissions or compliance, obtain authorized domain review before publication.

Failure 4: A one-time event becomes a permanent rule

An isolated observation turns into a universal policy. Review its effect on a real reader: would the person act too early, miss a valid interval, believe a process finished, or misplace an event? The repair should preserve both source-backed information and source uncertainty. Classify the failure as event order, aspect, reference anchor, recurrence, boundary, technical representation or evidence time. Compare an independent target-only interpretation with the approved source. If the difference affects safety, permissions or compliance, obtain authorized domain review before publication.

Failure 5: A forecast becomes a fact

Expected or may is replaced with did or will certainly. Review its effect on a real reader: would the person act too early, miss a valid interval, believe a process finished, or misplace an event? The repair should preserve both source-backed information and source uncertainty. Classify the failure as event order, aspect, reference anchor, recurrence, boundary, technical representation or evidence time. Compare an independent target-only interpretation with the approved source. If the difference affects safety, permissions or compliance, obtain authorized domain review before publication.

Failure 6: Quoted now moves to reader time

A historical speaker’s temporal perspective is lost. Review its effect on a real reader: would the person act too early, miss a valid interval, believe a process finished, or misplace an event? The repair should preserve both source-backed information and source uncertainty. Classify the failure as event order, aspect, reference anchor, recurrence, boundary, technical representation or evidence time. Compare an independent target-only interpretation with the approved source. If the difference affects safety, permissions or compliance, obtain authorized domain review before publication.

Failure 7: Two overlapping events become sequential

While or during is weakened into then. Review its effect on a real reader: would the person act too early, miss a valid interval, believe a process finished, or misplace an event? The repair should preserve both source-backed information and source uncertainty. Classify the failure as event order, aspect, reference anchor, recurrence, boundary, technical representation or evidence time. Compare an independent target-only interpretation with the approved source. If the difference affects safety, permissions or compliance, obtain authorized domain review before publication.

Failure 8: A required prerequisite becomes simultaneous

An action may appear permitted while its check is unfinished. Review its effect on a real reader: would the person act too early, miss a valid interval, believe a process finished, or misplace an event? The repair should preserve both source-backed information and source uncertainty. Classify the failure as event order, aspect, reference anchor, recurrence, boundary, technical representation or evidence time. Compare an independent target-only interpretation with the approved source. If the difference affects safety, permissions or compliance, obtain authorized domain review before publication.

Failure 9: After becomes because

Temporal order is made into an unsupported causal explanation. Review its effect on a real reader: would the person act too early, miss a valid interval, believe a process finished, or misplace an event? The repair should preserve both source-backed information and source uncertainty. Classify the failure as event order, aspect, reference anchor, recurrence, boundary, technical representation or evidence time. Compare an independent target-only interpretation with the approved source. If the difference affects safety, permissions or compliance, obtain authorized domain review before publication.

Failure 10: Before becomes by

A strict exclusion boundary changes into a deadline with different possible inclusivity. Review its effect on a real reader: would the person act too early, miss a valid interval, believe a process finished, or misplace an event? The repair should preserve both source-backed information and source uncertainty. Classify the failure as event order, aspect, reference anchor, recurrence, boundary, technical representation or evidence time. Compare an independent target-only interpretation with the approved source. If the difference affects safety, permissions or compliance, obtain authorized domain review before publication.

Failure 11: Until loses its stopping condition

An activity appears to stop before an authorized boundary. Review its effect on a real reader: would the person act too early, miss a valid interval, believe a process finished, or misplace an event? The repair should preserve both source-backed information and source uncertainty. Classify the failure as event order, aspect, reference anchor, recurrence, boundary, technical representation or evidence time. Compare an independent target-only interpretation with the approved source. If the difference affects safety, permissions or compliance, obtain authorized domain review before publication.

Failure 12: A duration becomes a clock time

Ninety minutes is treated as 01:30 without a start anchor. Review its effect on a real reader: would the person act too early, miss a valid interval, believe a process finished, or misplace an event? The repair should preserve both source-backed information and source uncertainty. Classify the failure as event order, aspect, reference anchor, recurrence, boundary, technical representation or evidence time. Compare an independent target-only interpretation with the approved source. If the difference affects safety, permissions or compliance, obtain authorized domain review before publication.

Failure 13: An approximate duration becomes exact

About five minutes turns into precisely 300 seconds. Review its effect on a real reader: would the person act too early, miss a valid interval, believe a process finished, or misplace an event? The repair should preserve both source-backed information and source uncertainty. Classify the failure as event order, aspect, reference anchor, recurrence, boundary, technical representation or evidence time. Compare an independent target-only interpretation with the approved source. If the difference affects safety, permissions or compliance, obtain authorized domain review before publication.

Failure 14: Calendar days become business days

Counting rules are changed without authority. Review its effect on a real reader: would the person act too early, miss a valid interval, believe a process finished, or misplace an event? The repair should preserve both source-backed information and source uncertainty. Classify the failure as event order, aspect, reference anchor, recurrence, boundary, technical representation or evidence time. Compare an independent target-only interpretation with the approved source. If the difference affects safety, permissions or compliance, obtain authorized domain review before publication.

Failure 15: Not yet becomes never

A temporary absence becomes permanent impossibility. Review its effect on a real reader: would the person act too early, miss a valid interval, believe a process finished, or misplace an event? The repair should preserve both source-backed information and source uncertainty. Classify the failure as event order, aspect, reference anchor, recurrence, boundary, technical representation or evidence time. Compare an independent target-only interpretation with the approved source. If the difference affects safety, permissions or compliance, obtain authorized domain review before publication.

Failure 16: No longer becomes not ever

A ceased state is reinterpreted as no past occurrence. Review its effect on a real reader: would the person act too early, miss a valid interval, believe a process finished, or misplace an event? The repair should preserve both source-backed information and source uncertainty. Classify the failure as event order, aspect, reference anchor, recurrence, boundary, technical representation or evidence time. Compare an independent target-only interpretation with the approved source. If the difference affects safety, permissions or compliance, obtain authorized domain review before publication.

Failure 17: A different time zone is silently used

A viewer sees a clock time for the wrong civil-time context. Review its effect on a real reader: would the person act too early, miss a valid interval, believe a process finished, or misplace an event? The repair should preserve both source-backed information and source uncertainty. Classify the failure as event order, aspect, reference anchor, recurrence, boundary, technical representation or evidence time. Compare an independent target-only interpretation with the approved source. If the difference affects safety, permissions or compliance, obtain authorized domain review before publication.

Failure 18: An offset becomes a timeless zone

Future recurrences drift across applicable offset changes. Review its effect on a real reader: would the person act too early, miss a valid interval, believe a process finished, or misplace an event? The repair should preserve both source-backed information and source uncertainty. Classify the failure as event order, aspect, reference anchor, recurrence, boundary, technical representation or evidence time. Compare an independent target-only interpretation with the approved source. If the difference affects safety, permissions or compliance, obtain authorized domain review before publication.

Failure 19: A daylight-saving overlap is unresolved

A repeated wall time is treated as one instant without additional data. Review its effect on a real reader: would the person act too early, miss a valid interval, believe a process finished, or misplace an event? The repair should preserve both source-backed information and source uncertainty. Classify the failure as event order, aspect, reference anchor, recurrence, boundary, technical representation or evidence time. Compare an independent target-only interpretation with the approved source. If the difference affects safety, permissions or compliance, obtain authorized domain review before publication.

Failure 20: A spring transition gap is ignored

A nonexistent civil time is treated as valid. Review its effect on a real reader: would the person act too early, miss a valid interval, believe a process finished, or misplace an event? The repair should preserve both source-backed information and source uncertainty. Classify the failure as event order, aspect, reference anchor, recurrence, boundary, technical representation or evidence time. Compare an independent target-only interpretation with the approved source. If the difference affects safety, permissions or compliance, obtain authorized domain review before publication.

Failure 21: An ambiguous numeric date is guessed

Day and month order is silently chosen. Review its effect on a real reader: would the person act too early, miss a valid interval, believe a process finished, or misplace an event? The repair should preserve both source-backed information and source uncertainty. Classify the failure as event order, aspect, reference anchor, recurrence, boundary, technical representation or evidence time. Compare an independent target-only interpretation with the approved source. If the difference affects safety, permissions or compliance, obtain authorized domain review before publication.

Failure 22: A date is stripped of its year

Archived and future dates become accidentally interchangeable. Review its effect on a real reader: would the person act too early, miss a valid interval, believe a process finished, or misplace an event? The repair should preserve both source-backed information and source uncertainty. Classify the failure as event order, aspect, reference anchor, recurrence, boundary, technical representation or evidence time. Compare an independent target-only interpretation with the approved source. If the difference affects safety, permissions or compliance, obtain authorized domain review before publication.

Failure 23: A relative UI phrase is cached

Tomorrow or recently remains static long after its anchor changes. Review its effect on a real reader: would the person act too early, miss a valid interval, believe a process finished, or misplace an event? The repair should preserve both source-backed information and source uncertainty. Classify the failure as event order, aspect, reference anchor, recurrence, boundary, technical representation or evidence time. Compare an independent target-only interpretation with the approved source. If the difference affects safety, permissions or compliance, obtain authorized domain review before publication.

Failure 24: Publication is confused with event occurrence

The document date replaces the event time. Review its effect on a real reader: would the person act too early, miss a valid interval, believe a process finished, or misplace an event? The repair should preserve both source-backed information and source uncertainty. Classify the failure as event order, aspect, reference anchor, recurrence, boundary, technical representation or evidence time. Compare an independent target-only interpretation with the approved source. If the difference affects safety, permissions or compliance, obtain authorized domain review before publication.

Failure 25: A later correction is backdated

Historical evidence is made to exist before its discovery. Review its effect on a real reader: would the person act too early, miss a valid interval, believe a process finished, or misplace an event? The repair should preserve both source-backed information and source uncertainty. Classify the failure as event order, aspect, reference anchor, recurrence, boundary, technical representation or evidence time. Compare an independent target-only interpretation with the approved source. If the difference affects safety, permissions or compliance, obtain authorized domain review before publication.

Failure 26: A conditional trigger is weakened

After approval becomes while awaiting approval. Review its effect on a real reader: would the person act too early, miss a valid interval, believe a process finished, or misplace an event? The repair should preserve both source-backed information and source uncertainty. Classify the failure as event order, aspect, reference anchor, recurrence, boundary, technical representation or evidence time. Compare an independent target-only interpretation with the approved source. If the difference affects safety, permissions or compliance, obtain authorized domain review before publication.

Failure 27: An unknown event order is fabricated

Unresolved testimony is made into confident chronology. Review its effect on a real reader: would the person act too early, miss a valid interval, believe a process finished, or misplace an event? The repair should preserve both source-backed information and source uncertainty. Classify the failure as event order, aspect, reference anchor, recurrence, boundary, technical representation or evidence time. Compare an independent target-only interpretation with the approved source. If the difference affects safety, permissions or compliance, obtain authorized domain review before publication.

Worked translation diagnostics: source, tempting mistake, repair and reader test

The following diagnostic exercises use English as an analysis language, not as a substitute for real target-language translation. Each source statement is paired with a tempting but incorrect paraphrase and a source-justified constraint. The translator’s job is to use target-language grammar and context to preserve the constraint. If the original source is materially ambiguous, clarify it with the information owner rather than guessing. The diagnostic tests tell reviewers what readers must be able to infer.

1. Earlier-than relations in past reports

The reference point is a later event viewed from within the past. The source may use a perfect construction or a time adverb; what matters is the earlier-than edge between the two events. A typical failure here is that the cause, preparation or completed condition may appear to follow the outcome. When analyzing any passage in this family, start with its source evidence, draw the minimal event graph, and decide which potential target-language interpretation must be ruled out. Independent reader testing matters because bilingual reviewers may unconsciously fill missing temporal information from a source text not visible to the final reader.

Case 1: The inspection had finished before the alert

Source claim. The expression “The inspection had finished before the alert” is a short example of earlier-than relations in past reports. An unsafe interpretation is “The inspection finished after the alert.” The source instead requires this relationship: Inspection ended earlier than alert. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. The reference point is a later event viewed from within the past. The source may use a perfect construction or a time adverb; what matters is the earlier-than edge between the two events. In this example the central risk is that the cause, preparation or completed condition may appear to follow the outcome. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “The inspection had finished before the alert” discusses. The RELATION and STATUS must support “Inspection ended earlier than alert” while refusing the unsupported alternative “The inspection finished after the alert.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask which event completed first. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Inspection ended earlier than alert” and does not casually assert “The inspection finished after the alert.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 2: She had already left when he called

Source claim. The expression “She had already left when he called” is a short example of earlier-than relations in past reports. An unsafe interpretation is “She left after answering his call.” The source instead requires this relationship: Departure preceded call. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. The reference point is a later event viewed from within the past. The source may use a perfect construction or a time adverb; what matters is the earlier-than edge between the two events. In this example the central risk is that the cause, preparation or completed condition may appear to follow the outcome. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “She had already left when he called” discusses. The RELATION and STATUS must support “Departure preceded call” while refusing the unsupported alternative “She left after answering his call.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Check whether she was present at call time. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Departure preceded call” and does not casually assert “She left after answering his call.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 3: The machines had cooled by the review

Source claim. The expression “The machines had cooled by the review” is a short example of earlier-than relations in past reports. An unsafe interpretation is “Review cooled the machines.” The source instead requires this relationship: Cooling was already complete at review. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. The reference point is a later event viewed from within the past. The source may use a perfect construction or a time adverb; what matters is the earlier-than edge between the two events. In this example the central risk is that the cause, preparation or completed condition may appear to follow the outcome. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “The machines had cooled by the review” discusses. The RELATION and STATUS must support “Cooling was already complete at review” while refusing the unsupported alternative “Review cooled the machines.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Test machine status when review started. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Cooling was already complete at review” and does not casually assert “Review cooled the machines.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 4: The archive had closed before the discovery

Source claim. The expression “The archive had closed before the discovery” is a short example of earlier-than relations in past reports. An unsafe interpretation is “Discovery caused closure.” The source instead requires this relationship: Closure occurred before discovery. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. The reference point is a later event viewed from within the past. The source may use a perfect construction or a time adverb; what matters is the earlier-than edge between the two events. In this example the central risk is that the cause, preparation or completed condition may appear to follow the outcome. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “The archive had closed before the discovery” discusses. The RELATION and STATUS must support “Closure occurred before discovery” while refusing the unsupported alternative “Discovery caused closure.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask if archive was open at discovery. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Closure occurred before discovery” and does not casually assert “Discovery caused closure.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 5: By dawn the team had crossed

Source claim. The expression “By dawn the team had crossed” is a short example of earlier-than relations in past reports. An unsafe interpretation is “Team started crossing at dawn.” The source instead requires this relationship: Crossing was completed before dawn. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. The reference point is a later event viewed from within the past. The source may use a perfect construction or a time adverb; what matters is the earlier-than edge between the two events. In this example the central risk is that the cause, preparation or completed condition may appear to follow the outcome. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “By dawn the team had crossed” discusses. The RELATION and STATUS must support “Crossing was completed before dawn” while refusing the unsupported alternative “Team started crossing at dawn.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask if crossing was ongoing at dawn. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Crossing was completed before dawn” and does not casually assert “Team started crossing at dawn.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 6: The permit had expired at the audit

Source claim. The expression “The permit had expired at the audit” is a short example of earlier-than relations in past reports. An unsafe interpretation is “Audit expired the permit.” The source instead requires this relationship: Expiry predates or coincides with audit reference. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. The reference point is a later event viewed from within the past. The source may use a perfect construction or a time adverb; what matters is the earlier-than edge between the two events. In this example the central risk is that the cause, preparation or completed condition may appear to follow the outcome. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “The permit had expired at the audit” discusses. The RELATION and STATUS must support “Expiry predates or coincides with audit reference” while refusing the unsupported alternative “Audit expired the permit.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Check validity during audit. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Expiry predates or coincides with audit reference” and does not casually assert “Audit expired the permit.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 7: She had sent the reply when meeting began

Source claim. The expression “She had sent the reply when meeting began” is a short example of earlier-than relations in past reports. An unsafe interpretation is “She sent it after the meeting.” The source instead requires this relationship: Reply was already sent at meeting start. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. The reference point is a later event viewed from within the past. The source may use a perfect construction or a time adverb; what matters is the earlier-than edge between the two events. In this example the central risk is that the cause, preparation or completed condition may appear to follow the outcome. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “She had sent the reply when meeting began” discusses. The RELATION and STATUS must support “Reply was already sent at meeting start” while refusing the unsupported alternative “She sent it after the meeting.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Test whether meeting participants could receive it. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Reply was already sent at meeting start” and does not casually assert “She sent it after the meeting.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 8: The storm had ended when the survey began

Source claim. The expression “The storm had ended when the survey began” is a short example of earlier-than relations in past reports. An unsafe interpretation is “Storm began during survey.” The source instead requires this relationship: Storm ended before survey. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. The reference point is a later event viewed from within the past. The source may use a perfect construction or a time adverb; what matters is the earlier-than edge between the two events. In this example the central risk is that the cause, preparation or completed condition may appear to follow the outcome. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “The storm had ended when the survey began” discusses. The RELATION and STATUS must support “Storm ended before survey” while refusing the unsupported alternative “Storm began during survey.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask whether the survey overlapped storm. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Storm ended before survey” and does not casually assert “Storm began during survey.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

2. Progressive activity versus completed results

An ongoing activity and an achieved result make different assertions. A perfect or progressive source expression may indicate work over time without establishing a completed deliverable. A typical failure here is that readers may treat pending or unfinished work as accepted, safe, delivered or final. When analyzing any passage in this family, start with its source evidence, draw the minimal event graph, and decide which potential target-language interpretation must be ruled out. Independent reader testing matters because bilingual reviewers may unconsciously fill missing temporal information from a source text not visible to the final reader.

Case 9: Researchers have been checking the figures

Source claim. The expression “Researchers have been checking the figures” is a short example of progressive activity versus completed results. An unsafe interpretation is “All figures are certified correct.” The source instead requires this relationship: Repeated checking does not prove approval. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. An ongoing activity and an achieved result make different assertions. A perfect or progressive source expression may indicate work over time without establishing a completed deliverable. In this example the central risk is that readers may treat pending or unfinished work as accepted, safe, delivered or final. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Researchers have been checking the figures” discusses. The RELATION and STATUS must support “Repeated checking does not prove approval” while refusing the unsupported alternative “All figures are certified correct.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask for signed verification. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Repeated checking does not prove approval” and does not casually assert “All figures are certified correct.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 10: The technician has repaired the unit

Source claim. The expression “The technician has repaired the unit” is a short example of progressive activity versus completed results. An unsafe interpretation is “The technician has been trying to repair it.” The source instead requires this relationship: Repair result is asserted. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. An ongoing activity and an achieved result make different assertions. A perfect or progressive source expression may indicate work over time without establishing a completed deliverable. In this example the central risk is that readers may treat pending or unfinished work as accepted, safe, delivered or final. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “The technician has repaired the unit” discusses. The RELATION and STATUS must support “Repair result is asserted” while refusing the unsupported alternative “The technician has been trying to repair it.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask if device is repaired. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Repair result is asserted” and does not casually assert “The technician has been trying to repair it.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 11: The board was still reviewing

Source claim. The expression “The board was still reviewing” is a short example of progressive activity versus completed results. An unsafe interpretation is “Board approved everything.” The source instead requires this relationship: Review was underway. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. An ongoing activity and an achieved result make different assertions. A perfect or progressive source expression may indicate work over time without establishing a completed deliverable. In this example the central risk is that readers may treat pending or unfinished work as accepted, safe, delivered or final. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “The board was still reviewing” discusses. The RELATION and STATUS must support “Review was underway” while refusing the unsupported alternative “Board approved everything.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask for resolution. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Review was underway” and does not casually assert “Board approved everything.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 12: The service is being restored

Source claim. The expression “The service is being restored” is a short example of progressive activity versus completed results. An unsafe interpretation is “Service has been restored.” The source instead requires this relationship: Restoration is unfinished or underway. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. An ongoing activity and an achieved result make different assertions. A perfect or progressive source expression may indicate work over time without establishing a completed deliverable. In this example the central risk is that readers may treat pending or unfinished work as accepted, safe, delivered or final. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “The service is being restored” discusses. The RELATION and STATUS must support “Restoration is unfinished or underway” while refusing the unsupported alternative “Service has been restored.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask if access is available. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Restoration is unfinished or underway” and does not casually assert “Service has been restored.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 13: She has completed the draft

Source claim. The expression “She has completed the draft” is a short example of progressive activity versus completed results. An unsafe interpretation is “She has started the draft.” The source instead requires this relationship: Draft endpoint is asserted. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. An ongoing activity and an achieved result make different assertions. A perfect or progressive source expression may indicate work over time without establishing a completed deliverable. In this example the central risk is that readers may treat pending or unfinished work as accepted, safe, delivered or final. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “She has completed the draft” discusses. The RELATION and STATUS must support “Draft endpoint is asserted” while refusing the unsupported alternative “She has started the draft.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Check submitted draft. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Draft endpoint is asserted” and does not casually assert “She has started the draft.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 14: They had begun but had not finished

Source claim. The expression “They had begun but had not finished” is a short example of progressive activity versus completed results. An unsafe interpretation is “They finished before starting.” The source instead requires this relationship: Start and finish are distinct events. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. An ongoing activity and an achieved result make different assertions. A perfect or progressive source expression may indicate work over time without establishing a completed deliverable. In this example the central risk is that readers may treat pending or unfinished work as accepted, safe, delivered or final. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “They had begun but had not finished” discusses. The RELATION and STATUS must support “Start and finish are distinct events” while refusing the unsupported alternative “They finished before starting.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask what remained. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Start and finish are distinct events” and does not casually assert “They finished before starting.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 15: She has lived there for years

Source claim. The expression “She has lived there for years” is a short example of progressive activity versus completed results. An unsafe interpretation is “She moved away years ago.” The source instead requires this relationship: An extended residence is claimed. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. An ongoing activity and an achieved result make different assertions. A perfect or progressive source expression may indicate work over time without establishing a completed deliverable. In this example the central risk is that readers may treat pending or unfinished work as accepted, safe, delivered or final. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “She has lived there for years” discusses. The RELATION and STATUS must support “An extended residence is claimed” while refusing the unsupported alternative “She moved away years ago.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask if current residence continues. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “An extended residence is claimed” and does not casually assert “She moved away years ago.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 16: The outage was being investigated

Source claim. The expression “The outage was being investigated” is a short example of progressive activity versus completed results. An unsafe interpretation is “Cause was confirmed.” The source instead requires this relationship: Investigation is not a verified finding. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. An ongoing activity and an achieved result make different assertions. A perfect or progressive source expression may indicate work over time without establishing a completed deliverable. In this example the central risk is that readers may treat pending or unfinished work as accepted, safe, delivered or final. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “The outage was being investigated” discusses. The RELATION and STATUS must support “Investigation is not a verified finding” while refusing the unsupported alternative “Cause was confirmed.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask if report exists. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Investigation is not a verified finding” and does not casually assert “Cause was confirmed.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

3. Habituality and recurring events

Habits describe regular patterns rather than one event. Frequency, exceptions, start periods and end conditions may be as important as tense. A typical failure here is that a recurring obligation may be treated as a one-off task or a conditional event may appear permanent. When analyzing any passage in this family, start with its source evidence, draw the minimal event graph, and decide which potential target-language interpretation must be ruled out. Independent reader testing matters because bilingual reviewers may unconsciously fill missing temporal information from a source text not visible to the final reader.

Case 17: The committee meets once a month

Source claim. The expression “The committee meets once a month” is a short example of habituality and recurring events. An unsafe interpretation is “It met only one time.” The source instead requires this relationship: Monthly recurrence. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Habits describe regular patterns rather than one event. Frequency, exceptions, start periods and end conditions may be as important as tense. In this example the central risk is that a recurring obligation may be treated as a one-off task or a conditional event may appear permanent. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “The committee meets once a month” discusses. The RELATION and STATUS must support “Monthly recurrence” while refusing the unsupported alternative “It met only one time.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Calculate three expected meetings. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Monthly recurrence” and does not casually assert “It met only one time.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 18: She used to inspect packages

Source claim. The expression “She used to inspect packages” is a short example of habituality and recurring events. An unsafe interpretation is “She currently inspects all packages.” The source instead requires this relationship: Past habit may have ended. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Habits describe regular patterns rather than one event. Frequency, exceptions, start periods and end conditions may be as important as tense. In this example the central risk is that a recurring obligation may be treated as a one-off task or a conditional event may appear permanent. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “She used to inspect packages” discusses. The RELATION and STATUS must support “Past habit may have ended” while refusing the unsupported alternative “She currently inspects all packages.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask if still true. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Past habit may have ended” and does not casually assert “She currently inspects all packages.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 19: The alarm occasionally repeats

Source claim. The expression “The alarm occasionally repeats” is a short example of habituality and recurring events. An unsafe interpretation is “Alarm always repeats.” The source instead requires this relationship: Low frequency, not universal. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Habits describe regular patterns rather than one event. Frequency, exceptions, start periods and end conditions may be as important as tense. In this example the central risk is that a recurring obligation may be treated as a one-off task or a conditional event may appear permanent. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “The alarm occasionally repeats” discusses. The RELATION and STATUS must support “Low frequency, not universal” while refusing the unsupported alternative “Alarm always repeats.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Check a no-repeat period. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Low frequency, not universal” and does not casually assert “Alarm always repeats.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 20: Every third run triggers review

Source claim. The expression “Every third run triggers review” is a short example of habituality and recurring events. An unsafe interpretation is “Each run triggers three reviews.” The source instead requires this relationship: Once per third run. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Habits describe regular patterns rather than one event. Frequency, exceptions, start periods and end conditions may be as important as tense. In this example the central risk is that a recurring obligation may be treated as a one-off task or a conditional event may appear permanent. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Every third run triggers review” discusses. The RELATION and STATUS must support “Once per third run” while refusing the unsupported alternative “Each run triggers three reviews.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Test cycles 3 and 6. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Once per third run” and does not casually assert “Each run triggers three reviews.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 21: Update is sent on working days

Source claim. The expression “Update is sent on working days” is a short example of habituality and recurring events. An unsafe interpretation is “Update sent all calendar days.” The source instead requires this relationship: Working-day schedule. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Habits describe regular patterns rather than one event. Frequency, exceptions, start periods and end conditions may be as important as tense. In this example the central risk is that a recurring obligation may be treated as a one-off task or a conditional event may appear permanent. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Update is sent on working days” discusses. The RELATION and STATUS must support “Working-day schedule” while refusing the unsupported alternative “Update sent all calendar days.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Test weekend. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Working-day schedule” and does not casually assert “Update sent all calendar days.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 22: Subscription renews each anniversary

Source claim. The expression “Subscription renews each anniversary” is a short example of habituality and recurring events. An unsafe interpretation is “It renews once.” The source instead requires this relationship: Annual recurrence. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Habits describe regular patterns rather than one event. Frequency, exceptions, start periods and end conditions may be as important as tense. In this example the central risk is that a recurring obligation may be treated as a one-off task or a conditional event may appear permanent. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Subscription renews each anniversary” discusses. The RELATION and STATUS must support “Annual recurrence” while refusing the unsupported alternative “It renews once.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Test second anniversary. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Annual recurrence” and does not casually assert “It renews once.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 23: Usually clears after a minute

Source claim. The expression “Usually clears after a minute” is a short example of habituality and recurring events. An unsafe interpretation is “Always clears in one minute.” The source instead requires this relationship: Typical behavior not guarantee. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Habits describe regular patterns rather than one event. Frequency, exceptions, start periods and end conditions may be as important as tense. In this example the central risk is that a recurring obligation may be treated as a one-off task or a conditional event may appear permanent. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Usually clears after a minute” discusses. The RELATION and STATUS must support “Typical behavior not guarantee” while refusing the unsupported alternative “Always clears in one minute.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Check longer episode. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Typical behavior not guarantee” and does not casually assert “Always clears in one minute.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 24: Maintenance excludes closure days

Source claim. The expression “Maintenance excludes closure days” is a short example of habituality and recurring events. An unsafe interpretation is “Maintenance has no exceptions.” The source instead requires this relationship: Explicit exception. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Habits describe regular patterns rather than one event. Frequency, exceptions, start periods and end conditions may be as important as tense. In this example the central risk is that a recurring obligation may be treated as a one-off task or a conditional event may appear permanent. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Maintenance excludes closure days” discusses. The RELATION and STATUS must support “Explicit exception” while refusing the unsupported alternative “Maintenance has no exceptions.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Check closure date. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Explicit exception” and does not casually assert “Maintenance has no exceptions.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

4. Overlap and simultaneous actions

When, while and during can establish an overlap relation, not merely chronological sequence. A source may also describe background activity interrupted by a shorter event. A typical failure here is that a reader may infer one action happened only after another finished. When analyzing any passage in this family, start with its source evidence, draw the minimal event graph, and decide which potential target-language interpretation must be ruled out. Independent reader testing matters because bilingual reviewers may unconsciously fill missing temporal information from a source text not visible to the final reader.

Case 25: She checked logs while uploading

Source claim. The expression “She checked logs while uploading” is a short example of overlap and simultaneous actions. An unsafe interpretation is “She checked only after upload.” The source instead requires this relationship: Tasks overlapped. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. When, while and during can establish an overlap relation, not merely chronological sequence. A source may also describe background activity interrupted by a shorter event. In this example the central risk is that a reader may infer one action happened only after another finished. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “She checked logs while uploading” discusses. The RELATION and STATUS must support “Tasks overlapped” while refusing the unsupported alternative “She checked only after upload.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask upload state during checking. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Tasks overlapped” and does not casually assert “She checked only after upload.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 26: The alarm sounded during rehearsal

Source claim. The expression “The alarm sounded during rehearsal” is a short example of overlap and simultaneous actions. An unsafe interpretation is “Alarm came before rehearsal.” The source instead requires this relationship: Alarm inside rehearsal interval. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. When, while and during can establish an overlap relation, not merely chronological sequence. A source may also describe background activity interrupted by a shorter event. In this example the central risk is that a reader may infer one action happened only after another finished. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “The alarm sounded during rehearsal” discusses. The RELATION and STATUS must support “Alarm inside rehearsal interval” while refusing the unsupported alternative “Alarm came before rehearsal.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Locate rehearsal boundaries. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Alarm inside rehearsal interval” and does not casually assert “Alarm came before rehearsal.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 27: The system was cooling when power failed

Source claim. The expression “The system was cooling when power failed” is a short example of overlap and simultaneous actions. An unsafe interpretation is “Cooling began after failure.” The source instead requires this relationship: Cooling preceded and overlapped failure. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. When, while and during can establish an overlap relation, not merely chronological sequence. A source may also describe background activity interrupted by a shorter event. In this example the central risk is that a reader may infer one action happened only after another finished. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “The system was cooling when power failed” discusses. The RELATION and STATUS must support “Cooling preceded and overlapped failure” while refusing the unsupported alternative “Cooling began after failure.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask state just before failure. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Cooling preceded and overlapped failure” and does not casually assert “Cooling began after failure.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 28: He took notes as she explained

Source claim. The expression “He took notes as she explained” is a short example of overlap and simultaneous actions. An unsafe interpretation is “He took notes afterward.” The source instead requires this relationship: Simultaneous activities. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. When, while and during can establish an overlap relation, not merely chronological sequence. A source may also describe background activity interrupted by a shorter event. In this example the central risk is that a reader may infer one action happened only after another finished. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “He took notes as she explained” discusses. The RELATION and STATUS must support “Simultaneous activities” while refusing the unsupported alternative “He took notes afterward.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask if notes arose during explanation. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Simultaneous activities” and does not casually assert “He took notes afterward.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 29: As the door opened lights flickered

Source claim. The expression “As the door opened lights flickered” is a short example of overlap and simultaneous actions. An unsafe interpretation is “Lights flickered long afterward.” The source instead requires this relationship: Closely overlapping onsets. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. When, while and during can establish an overlap relation, not merely chronological sequence. A source may also describe background activity interrupted by a shorter event. In this example the central risk is that a reader may infer one action happened only after another finished. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “As the door opened lights flickered” discusses. The RELATION and STATUS must support “Closely overlapping onsets” while refusing the unsupported alternative “Lights flickered long afterward.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask for event gap. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Closely overlapping onsets” and does not casually assert “Lights flickered long afterward.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 30: The recording continued while he slept

Source claim. The expression “The recording continued while he slept” is a short example of overlap and simultaneous actions. An unsafe interpretation is “Recording stopped as he slept.” The source instead requires this relationship: Recording overlapped sleep. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. When, while and during can establish an overlap relation, not merely chronological sequence. A source may also describe background activity interrupted by a shorter event. In this example the central risk is that a reader may infer one action happened only after another finished. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “The recording continued while he slept” discusses. The RELATION and STATUS must support “Recording overlapped sleep” while refusing the unsupported alternative “Recording stopped as he slept.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Check recording status. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Recording overlapped sleep” and does not casually assert “Recording stopped as he slept.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 31: The alert interrupted the test

Source claim. The expression “The alert interrupted the test” is a short example of overlap and simultaneous actions. An unsafe interpretation is “Alert came after test ended.” The source instead requires this relationship: Test was in progress at alert. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. When, while and during can establish an overlap relation, not merely chronological sequence. A source may also describe background activity interrupted by a shorter event. In this example the central risk is that a reader may infer one action happened only after another finished. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “The alert interrupted the test” discusses. The RELATION and STATUS must support “Test was in progress at alert” while refusing the unsupported alternative “Alert came after test ended.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask if test finished. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Test was in progress at alert” and does not casually assert “Alert came after test ended.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 32: The queue grew as capacity stayed constant

Source claim. The expression “The queue grew as capacity stayed constant” is a short example of overlap and simultaneous actions. An unsafe interpretation is “Capacity grew with queue.” The source instead requires this relationship: Growth overlaps fixed capacity. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. When, while and during can establish an overlap relation, not merely chronological sequence. A source may also describe background activity interrupted by a shorter event. In this example the central risk is that a reader may infer one action happened only after another finished. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “The queue grew as capacity stayed constant” discusses. The RELATION and STATUS must support “Growth overlaps fixed capacity” while refusing the unsupported alternative “Capacity grew with queue.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask capacity during growth. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Growth overlaps fixed capacity” and does not casually assert “Capacity grew with queue.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

5. Procedure order, approvals and causes

Procedural sequencing grants or withholds permission. A request for approval is not the same event as approval, and temporal succession does not by itself establish causation. A typical failure here is that the translated action may become permitted too early, or a mere sequence may be misreported as a cause. When analyzing any passage in this family, start with its source evidence, draw the minimal event graph, and decide which potential target-language interpretation must be ruled out. Independent reader testing matters because bilingual reviewers may unconsciously fill missing temporal information from a source text not visible to the final reader.

Case 33: Verify identity before opening record

Source claim. The expression “Verify identity before opening record” is a short example of procedure order, approvals and causes. An unsafe interpretation is “Open then verify.” The source instead requires this relationship: Verification is prerequisite. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Procedural sequencing grants or withholds permission. A request for approval is not the same event as approval, and temporal succession does not by itself establish causation. In this example the central risk is that the translated action may become permitted too early, or a mere sequence may be misreported as a cause. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Verify identity before opening record” discusses. The RELATION and STATUS must support “Verification is prerequisite” while refusing the unsupported alternative “Open then verify.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Try without identity. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Verification is prerequisite” and does not casually assert “Open then verify.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 34: Release after approval

Source claim. The expression “Release after approval” is a short example of procedure order, approvals and causes. An unsafe interpretation is “Release when approval requested.” The source instead requires this relationship: Approval must actually occur. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Procedural sequencing grants or withholds permission. A request for approval is not the same event as approval, and temporal succession does not by itself establish causation. In this example the central risk is that the translated action may become permitted too early, or a mere sequence may be misreported as a cause. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Release after approval” discusses. The RELATION and STATUS must support “Approval must actually occur” while refusing the unsupported alternative “Release when approval requested.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Test pending approval. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Approval must actually occur” and does not casually assert “Release when approval requested.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 35: Do not restart until backup completes

Source claim. The expression “Do not restart until backup completes” is a short example of procedure order, approvals and causes. An unsafe interpretation is “Restart while backup runs.” The source instead requires this relationship: Backup must finish first. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Procedural sequencing grants or withholds permission. A request for approval is not the same event as approval, and temporal succession does not by itself establish causation. In this example the central risk is that the translated action may become permitted too early, or a mere sequence may be misreported as a cause. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Do not restart until backup completes” discusses. The RELATION and STATUS must support “Backup must finish first” while refusing the unsupported alternative “Restart while backup runs.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Test in-progress backup. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Backup must finish first” and does not casually assert “Restart while backup runs.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 36: Save draft then request sign-off

Source claim. The expression “Save draft then request sign-off” is a short example of procedure order, approvals and causes. An unsafe interpretation is “Request sign-off before draft saved.” The source instead requires this relationship: Saving precedes request. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Procedural sequencing grants or withholds permission. A request for approval is not the same event as approval, and temporal succession does not by itself establish causation. In this example the central risk is that the translated action may become permitted too early, or a mere sequence may be misreported as a cause. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Save draft then request sign-off” discusses. The RELATION and STATUS must support “Saving precedes request” while refusing the unsupported alternative “Request sign-off before draft saved.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Check available draft. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Saving precedes request” and does not casually assert “Request sign-off before draft saved.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 37: Weigh once the sample cools

Source claim. The expression “Weigh once the sample cools” is a short example of procedure order, approvals and causes. An unsafe interpretation is “Weigh during cooling.” The source instead requires this relationship: Cooling state qualifies weighing. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Procedural sequencing grants or withholds permission. A request for approval is not the same event as approval, and temporal succession does not by itself establish causation. In this example the central risk is that the translated action may become permitted too early, or a mere sequence may be misreported as a cause. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Weigh once the sample cools” discusses. The RELATION and STATUS must support “Cooling state qualifies weighing” while refusing the unsupported alternative “Weigh during cooling.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Test warm sample. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Cooling state qualifies weighing” and does not casually assert “Weigh during cooling.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 38: Alarm sounded after door closed

Source claim. The expression “Alarm sounded after door closed” is a short example of procedure order, approvals and causes. An unsafe interpretation is “Alarm caused door to close.” The source instead requires this relationship: Temporal relation not causal proof. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Procedural sequencing grants or withholds permission. A request for approval is not the same event as approval, and temporal succession does not by itself establish causation. In this example the central risk is that the translated action may become permitted too early, or a mere sequence may be misreported as a cause. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Alarm sounded after door closed” discusses. The RELATION and STATUS must support “Temporal relation not causal proof” while refusing the unsupported alternative “Alarm caused door to close.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask if causation asserted. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Temporal relation not causal proof” and does not casually assert “Alarm caused door to close.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 39: Correct error before retesting

Source claim. The expression “Correct error before retesting” is a short example of procedure order, approvals and causes. An unsafe interpretation is “Retest before correction.” The source instead requires this relationship: Correction precedes retest. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Procedural sequencing grants or withholds permission. A request for approval is not the same event as approval, and temporal succession does not by itself establish causation. In this example the central risk is that the translated action may become permitted too early, or a mere sequence may be misreported as a cause. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Correct error before retesting” discusses. The RELATION and STATUS must support “Correction precedes retest” while refusing the unsupported alternative “Retest before correction.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Test failure workflow. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Correction precedes retest” and does not casually assert “Retest before correction.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 40: Publish results after decision recorded

Source claim. The expression “Publish results after decision recorded” is a short example of procedure order, approvals and causes. An unsafe interpretation is “Publish before decision recorded.” The source instead requires this relationship: Record must precede publication. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Procedural sequencing grants or withholds permission. A request for approval is not the same event as approval, and temporal succession does not by itself establish causation. In this example the central risk is that the translated action may become permitted too early, or a mere sequence may be misreported as a cause. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Publish results after decision recorded” discusses. The RELATION and STATUS must support “Record must precede publication” while refusing the unsupported alternative “Publish before decision recorded.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask for decision record. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Record must precede publication” and does not casually assert “Publish before decision recorded.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

6. Reported speech and old relative words

A quotation carries its own speech time, and reported speech may shift grammatical viewpoint. Tomorrow, now and ago in a historical quotation are not anchored to a modern reading date. A typical failure here is that past promises may seem live, and past observations may become present status claims. When analyzing any passage in this family, start with its source evidence, draw the minimal event graph, and decide which potential target-language interpretation must be ruled out. Independent reader testing matters because bilingual reviewers may unconsciously fill missing temporal information from a source text not visible to the final reader.

Case 41: On Tuesday she said I will call tomorrow

Source claim. The expression “On Tuesday she said I will call tomorrow” is a short example of reported speech and old relative words. An unsafe interpretation is “Call is tomorrow after translation.” The source instead requires this relationship: Tomorrow anchored to Tuesday. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. A quotation carries its own speech time, and reported speech may shift grammatical viewpoint. Tomorrow, now and ago in a historical quotation are not anchored to a modern reading date. In this example the central risk is that past promises may seem live, and past observations may become present status claims. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “On Tuesday she said I will call tomorrow” discusses. The RELATION and STATUS must support “Tomorrow anchored to Tuesday” while refusing the unsupported alternative “Call is tomorrow after translation.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask intended Wednesday. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Tomorrow anchored to Tuesday” and does not casually assert “Call is tomorrow after translation.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 42: Old notice says the system is down now

Source claim. The expression “Old notice says the system is down now” is a short example of reported speech and old relative words. An unsafe interpretation is “System is down this instant.” The source instead requires this relationship: Now belongs to notice date. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. A quotation carries its own speech time, and reported speech may shift grammatical viewpoint. Tomorrow, now and ago in a historical quotation are not anchored to a modern reading date. In this example the central risk is that past promises may seem live, and past observations may become present status claims. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Old notice says the system is down now” discusses. The RELATION and STATUS must support “Now belongs to notice date” while refusing the unsupported alternative “System is down this instant.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask current status. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Now belongs to notice date” and does not casually assert “System is down this instant.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 43: Witness says gate had been open earlier

Source claim. The expression “Witness says gate had been open earlier” is a short example of reported speech and old relative words. An unsafe interpretation is “Gate opened during testimony.” The source instead requires this relationship: Open state preceded witness report. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. A quotation carries its own speech time, and reported speech may shift grammatical viewpoint. Tomorrow, now and ago in a historical quotation are not anchored to a modern reading date. In this example the central risk is that past promises may seem live, and past observations may become present status claims. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Witness says gate had been open earlier” discusses. The RELATION and STATUS must support “Open state preceded witness report” while refusing the unsupported alternative “Gate opened during testimony.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask earlier period. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Open state preceded witness report” and does not casually assert “Gate opened during testimony.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 44: He recalled he had never met them

Source claim. The expression “He recalled he had never met them” is a short example of reported speech and old relative words. An unsafe interpretation is “He vowed never to meet them.” The source instead requires this relationship: Never limits past experience. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. A quotation carries its own speech time, and reported speech may shift grammatical viewpoint. Tomorrow, now and ago in a historical quotation are not anchored to a modern reading date. In this example the central risk is that past promises may seem live, and past observations may become present status claims. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “He recalled he had never met them” discusses. The RELATION and STATUS must support “Never limits past experience” while refusing the unsupported alternative “He vowed never to meet them.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask about future possibility. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Never limits past experience” and does not casually assert “He vowed never to meet them.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 45: She said she would depart Friday

Source claim. The expression “She said she would depart Friday” is a short example of reported speech and old relative words. An unsafe interpretation is “She certainly departed Friday.” The source instead requires this relationship: Past expectation versus actual departure. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. A quotation carries its own speech time, and reported speech may shift grammatical viewpoint. Tomorrow, now and ago in a historical quotation are not anchored to a modern reading date. In this example the central risk is that past promises may seem live, and past observations may become present status claims. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “She said she would depart Friday” discusses. The RELATION and STATUS must support “Past expectation versus actual departure” while refusing the unsupported alternative “She certainly departed Friday.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Check journey record. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Past expectation versus actual departure” and does not casually assert “She certainly departed Friday.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 46: We arrived two hours ago in transcript

Source claim. The expression “We arrived two hours ago in transcript” is a short example of reported speech and old relative words. An unsafe interpretation is “We arrive two hours later.” The source instead requires this relationship: Ago points before speech time. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. A quotation carries its own speech time, and reported speech may shift grammatical viewpoint. Tomorrow, now and ago in a historical quotation are not anchored to a modern reading date. In this example the central risk is that past promises may seem live, and past observations may become present status claims. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “We arrived two hours ago in transcript” discusses. The RELATION and STATUS must support “Ago points before speech time” while refusing the unsupported alternative “We arrive two hours later.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Compute arrival time. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Ago points before speech time” and does not casually assert “We arrive two hours later.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 47: Journal calls this evening quiet

Source claim. The expression “Journal calls this evening quiet” is a short example of reported speech and old relative words. An unsafe interpretation is “Current evening is quiet.” The source instead requires this relationship: Evening belongs to dated journal. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. A quotation carries its own speech time, and reported speech may shift grammatical viewpoint. Tomorrow, now and ago in a historical quotation are not anchored to a modern reading date. In this example the central risk is that past promises may seem live, and past observations may become present status claims. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Journal calls this evening quiet” discusses. The RELATION and STATUS must support “Evening belongs to dated journal” while refusing the unsupported alternative “Current evening is quiet.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Find entry date. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Evening belongs to dated journal” and does not casually assert “Current evening is quiet.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 48: Warning said event might occur next day

Source claim. The expression “Warning said event might occur next day” is a short example of reported speech and old relative words. An unsafe interpretation is “Event occurred next day.” The source instead requires this relationship: Warning is uncertain future statement. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. A quotation carries its own speech time, and reported speech may shift grammatical viewpoint. Tomorrow, now and ago in a historical quotation are not anchored to a modern reading date. In this example the central risk is that past promises may seem live, and past observations may become present status claims. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Warning said event might occur next day” discusses. The RELATION and STATUS must support “Warning is uncertain future statement” while refusing the unsupported alternative “Event occurred next day.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask for outcome. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Warning is uncertain future statement” and does not casually assert “Event occurred next day.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

7. Future-in-the-past and uncertain outcomes

Expected, hoped, was to and would represent future from an earlier vantage point, not necessarily actual fulfillment. Modal strength belongs to temporal meaning. A typical failure here is that a planned or merely possible event may become a confirmed historical fact. When analyzing any passage in this family, start with its source evidence, draw the minimal event graph, and decide which potential target-language interpretation must be ruled out. Independent reader testing matters because bilingual reviewers may unconsciously fill missing temporal information from a source text not visible to the final reader.

Case 49: They expected launch by June

Source claim. The expression “They expected launch by June” is a short example of future-in-the-past and uncertain outcomes. An unsafe interpretation is “Launch happened by June.” The source instead requires this relationship: Expectation, not achieved result. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Expected, hoped, was to and would represent future from an earlier vantage point, not necessarily actual fulfillment. Modal strength belongs to temporal meaning. In this example the central risk is that a planned or merely possible event may become a confirmed historical fact. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “They expected launch by June” discusses. The RELATION and STATUS must support “Expectation, not achieved result” while refusing the unsupported alternative “Launch happened by June.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Check launch record. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Expectation, not achieved result” and does not casually assert “Launch happened by June.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 50: She was going to call

Source claim. The expression “She was going to call” is a short example of future-in-the-past and uncertain outcomes. An unsafe interpretation is “She called.” The source instead requires this relationship: Intention is not outcome. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Expected, hoped, was to and would represent future from an earlier vantage point, not necessarily actual fulfillment. Modal strength belongs to temporal meaning. In this example the central risk is that a planned or merely possible event may become a confirmed historical fact. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “She was going to call” discusses. The RELATION and STATUS must support “Intention is not outcome” while refusing the unsupported alternative “She called.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask for call record. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Intention is not outcome” and does not casually assert “She called.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 51: Forecast predicted demand increase

Source claim. The expression “Forecast predicted demand increase” is a short example of future-in-the-past and uncertain outcomes. An unsafe interpretation is “Demand definitely rose.” The source instead requires this relationship: Forecast is prospective. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Expected, hoped, was to and would represent future from an earlier vantage point, not necessarily actual fulfillment. Modal strength belongs to temporal meaning. In this example the central risk is that a planned or merely possible event may become a confirmed historical fact. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Forecast predicted demand increase” discusses. The RELATION and STATUS must support “Forecast is prospective” while refusing the unsupported alternative “Demand definitely rose.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Check measured data. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Forecast is prospective” and does not casually assert “Demand definitely rose.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 52: He feared failure

Source claim. The expression “He feared failure” is a short example of future-in-the-past and uncertain outcomes. An unsafe interpretation is “Failure occurred.” The source instead requires this relationship: Fear not confirmation. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Expected, hoped, was to and would represent future from an earlier vantage point, not necessarily actual fulfillment. Modal strength belongs to temporal meaning. In this example the central risk is that a planned or merely possible event may become a confirmed historical fact. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “He feared failure” discusses. The RELATION and STATUS must support “Fear not confirmation” while refusing the unsupported alternative “Failure occurred.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask if device failed. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Fear not confirmation” and does not casually assert “Failure occurred.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 53: Schedule showed Monday start

Source claim. The expression “Schedule showed Monday start” is a short example of future-in-the-past and uncertain outcomes. An unsafe interpretation is “Activity started Monday.” The source instead requires this relationship: Planned date not necessarily actual. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Expected, hoped, was to and would represent future from an earlier vantage point, not necessarily actual fulfillment. Modal strength belongs to temporal meaning. In this example the central risk is that a planned or merely possible event may become a confirmed historical fact. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Schedule showed Monday start” discusses. The RELATION and STATUS must support “Planned date not necessarily actual” while refusing the unsupported alternative “Activity started Monday.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Check start log. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Planned date not necessarily actual” and does not casually assert “Activity started Monday.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 54: Estimated delay might increase

Source claim. The expression “Estimated delay might increase” is a short example of future-in-the-past and uncertain outcomes. An unsafe interpretation is “Delay surely increased.” The source instead requires this relationship: Uncertain forecast. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Expected, hoped, was to and would represent future from an earlier vantage point, not necessarily actual fulfillment. Modal strength belongs to temporal meaning. In this example the central risk is that a planned or merely possible event may become a confirmed historical fact. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Estimated delay might increase” discusses. The RELATION and STATUS must support “Uncertain forecast” while refusing the unsupported alternative “Delay surely increased.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Check observed delay. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Uncertain forecast” and does not casually assert “Delay surely increased.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 55: Successor would be appointed later

Source claim. The expression “Successor would be appointed later” is a short example of future-in-the-past and uncertain outcomes. An unsafe interpretation is “Successor already appointed.” The source instead requires this relationship: Future relative to document. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Expected, hoped, was to and would represent future from an earlier vantage point, not necessarily actual fulfillment. Modal strength belongs to temporal meaning. In this example the central risk is that a planned or merely possible event may become a confirmed historical fact. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Successor would be appointed later” discusses. The RELATION and STATUS must support “Future relative to document” while refusing the unsupported alternative “Successor already appointed.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Check announcement. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Future relative to document” and does not casually assert “Successor already appointed.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 56: Team hoped to repeat trial

Source claim. The expression “Team hoped to repeat trial” is a short example of future-in-the-past and uncertain outcomes. An unsafe interpretation is “Second trial took place.” The source instead requires this relationship: Hope is not action. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Expected, hoped, was to and would represent future from an earlier vantage point, not necessarily actual fulfillment. Modal strength belongs to temporal meaning. In this example the central risk is that a planned or merely possible event may become a confirmed historical fact. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Team hoped to repeat trial” discusses. The RELATION and STATUS must support “Hope is not action” while refusing the unsupported alternative “Second trial took place.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Check new trial record. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Hope is not action” and does not casually assert “Second trial took place.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

8. Not yet, no longer and temporal negatives

Not yet describes absence at the reference point without ruling out a later event. No longer indicates cessation, while never can exclude the whole relevant past interval. A typical failure here is that a temporary absence becomes a permanent condition or an ended habit becomes no past occurrence. When analyzing any passage in this family, start with its source evidence, draw the minimal event graph, and decide which potential target-language interpretation must be ruled out. Independent reader testing matters because bilingual reviewers may unconsciously fill missing temporal information from a source text not visible to the final reader.

Case 57: Results are not yet available

Source claim. The expression “Results are not yet available” is a short example of not yet, no longer and temporal negatives. An unsafe interpretation is “Results will never be available.” The source instead requires this relationship: Future availability remains possible. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Not yet describes absence at the reference point without ruling out a later event. No longer indicates cessation, while never can exclude the whole relevant past interval. In this example the central risk is that a temporary absence becomes a permanent condition or an ended habit becomes no past occurrence. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Results are not yet available” discusses. The RELATION and STATUS must support “Future availability remains possible” while refusing the unsupported alternative “Results will never be available.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask about next release. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Future availability remains possible” and does not casually assert “Results will never be available.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 58: Sensor no longer responds

Source claim. The expression “Sensor no longer responds” is a short example of not yet, no longer and temporal negatives. An unsafe interpretation is “Sensor never responded.” The source instead requires this relationship: Earlier responsiveness may exist. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Not yet describes absence at the reference point without ruling out a later event. No longer indicates cessation, while never can exclude the whole relevant past interval. In this example the central risk is that a temporary absence becomes a permanent condition or an ended habit becomes no past occurrence. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Sensor no longer responds” discusses. The RELATION and STATUS must support “Earlier responsiveness may exist” while refusing the unsupported alternative “Sensor never responded.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Check past readings. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Earlier responsiveness may exist” and does not casually assert “Sensor never responded.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 59: No update arrived by noon

Source claim. The expression “No update arrived by noon” is a short example of not yet, no longer and temporal negatives. An unsafe interpretation is “No update ever arrived.” The source instead requires this relationship: Absence is limited to noon. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Not yet describes absence at the reference point without ruling out a later event. No longer indicates cessation, while never can exclude the whole relevant past interval. In this example the central risk is that a temporary absence becomes a permanent condition or an ended habit becomes no past occurrence. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “No update arrived by noon” discusses. The RELATION and STATUS must support “Absence is limited to noon” while refusing the unsupported alternative “No update ever arrived.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Check afternoon. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Absence is limited to noon” and does not casually assert “No update ever arrived.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 60: Alert stopped repeating

Source claim. The expression “Alert stopped repeating” is a short example of not yet, no longer and temporal negatives. An unsafe interpretation is “Alert did not repeat before.” The source instead requires this relationship: Repetition existed then ceased. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Not yet describes absence at the reference point without ruling out a later event. No longer indicates cessation, while never can exclude the whole relevant past interval. In this example the central risk is that a temporary absence becomes a permanent condition or an ended habit becomes no past occurrence. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Alert stopped repeating” discusses. The RELATION and STATUS must support “Repetition existed then ceased” while refusing the unsupported alternative “Alert did not repeat before.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Check earlier signals. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Repetition existed then ceased” and does not casually assert “Alert did not repeat before.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 61: He has not begun work

Source claim. The expression “He has not begun work” is a short example of not yet, no longer and temporal negatives. An unsafe interpretation is “He finished work.” The source instead requires this relationship: No start, not completed end. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Not yet describes absence at the reference point without ruling out a later event. No longer indicates cessation, while never can exclude the whole relevant past interval. In this example the central risk is that a temporary absence becomes a permanent condition or an ended habit becomes no past occurrence. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “He has not begun work” discusses. The RELATION and STATUS must support “No start, not completed end” while refusing the unsupported alternative “He finished work.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask about start. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “No start, not completed end” and does not casually assert “He finished work.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 62: Service was not always active

Source claim. The expression “Service was not always active” is a short example of not yet, no longer and temporal negatives. An unsafe interpretation is “Service was never active.” The source instead requires this relationship: Some inactive intervals only. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Not yet describes absence at the reference point without ruling out a later event. No longer indicates cessation, while never can exclude the whole relevant past interval. In this example the central risk is that a temporary absence becomes a permanent condition or an ended habit becomes no past occurrence. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Service was not always active” discusses. The RELATION and STATUS must support “Some inactive intervals only” while refusing the unsupported alternative “Service was never active.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Check other periods. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Some inactive intervals only” and does not casually assert “Service was never active.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 63: Restriction is temporary

Source claim. The expression “Restriction is temporary” is a short example of not yet, no longer and temporal negatives. An unsafe interpretation is “Restriction is permanent.” The source instead requires this relationship: Limited period. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Not yet describes absence at the reference point without ruling out a later event. No longer indicates cessation, while never can exclude the whole relevant past interval. In this example the central risk is that a temporary absence becomes a permanent condition or an ended habit becomes no past occurrence. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Restriction is temporary” discusses. The RELATION and STATUS must support “Limited period” while refusing the unsupported alternative “Restriction is permanent.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask expiry rule. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Limited period” and does not casually assert “Restriction is permanent.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 64: Change not effective until approval

Source claim. The expression “Change not effective until approval” is a short example of not yet, no longer and temporal negatives. An unsafe interpretation is “Change effective during review.” The source instead requires this relationship: Approval controls effectivity. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Not yet describes absence at the reference point without ruling out a later event. No longer indicates cessation, while never can exclude the whole relevant past interval. In this example the central risk is that a temporary absence becomes a permanent condition or an ended habit becomes no past occurrence. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Change not effective until approval” discusses. The RELATION and STATUS must support “Approval controls effectivity” while refusing the unsupported alternative “Change effective during review.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Check authorization. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Approval controls effectivity” and does not casually assert “Change effective during review.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

9. Deadlines, exclusion and inclusion

Before, by, through, until and within define intervals or constraints whose endpoints may differ. The governing source must resolve precise policy boundaries. A typical failure here is that eligible actions can be rejected, forbidden actions permitted or users notified on the wrong date. When analyzing any passage in this family, start with its source evidence, draw the minimal event graph, and decide which potential target-language interpretation must be ruled out. Independent reader testing matters because bilingual reviewers may unconsciously fill missing temporal information from a source text not visible to the final reader.

Case 65: Submit before 17:00

Source claim. The expression “Submit before 17:00” is a short example of deadlines, exclusion and inclusion. An unsafe interpretation is “17:00 itself accepted.” The source instead requires this relationship: Strict earlier-than. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Before, by, through, until and within define intervals or constraints whose endpoints may differ. The governing source must resolve precise policy boundaries. In this example the central risk is that eligible actions can be rejected, forbidden actions permitted or users notified on the wrong date. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Submit before 17:00” discusses. The RELATION and STATUS must support “Strict earlier-than” while refusing the unsupported alternative “17:00 itself accepted.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Test exact 17:00. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Strict earlier-than” and does not casually assert “17:00 itself accepted.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 66: Submit by 17:00

Source claim. The expression “Submit by 17:00” is a short example of deadlines, exclusion and inclusion. An unsafe interpretation is “Submit after 17:00.” The source instead requires this relationship: Latest permissible time. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Before, by, through, until and within define intervals or constraints whose endpoints may differ. The governing source must resolve precise policy boundaries. In this example the central risk is that eligible actions can be rejected, forbidden actions permitted or users notified on the wrong date. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Submit by 17:00” discusses. The RELATION and STATUS must support “Latest permissible time” while refusing the unsupported alternative “Submit after 17:00.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Test earlier and later. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Latest permissible time” and does not casually assert “Submit after 17:00.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 67: Offer runs through Friday

Source claim. The expression “Offer runs through Friday” is a short example of deadlines, exclusion and inclusion. An unsafe interpretation is “Offer ends Thursday.” The source instead requires this relationship: Friday included in ordinary phrasing. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Before, by, through, until and within define intervals or constraints whose endpoints may differ. The governing source must resolve precise policy boundaries. In this example the central risk is that eligible actions can be rejected, forbidden actions permitted or users notified on the wrong date. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Offer runs through Friday” discusses. The RELATION and STATUS must support “Friday included in ordinary phrasing” while refusing the unsupported alternative “Offer ends Thursday.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Test Friday. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Friday included in ordinary phrasing” and does not casually assert “Offer ends Thursday.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 68: Do not enter until approved

Source claim. The expression “Do not enter until approved” is a short example of deadlines, exclusion and inclusion. An unsafe interpretation is “Enter during approval.” The source instead requires this relationship: Approval gate. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Before, by, through, until and within define intervals or constraints whose endpoints may differ. The governing source must resolve precise policy boundaries. In this example the central risk is that eligible actions can be rejected, forbidden actions permitted or users notified on the wrong date. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Do not enter until approved” discusses. The RELATION and STATUS must support “Approval gate” while refusing the unsupported alternative “Enter during approval.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Test pending status. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Approval gate” and does not casually assert “Enter during approval.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 69: Applications open Monday

Source claim. The expression “Applications open Monday” is a short example of deadlines, exclusion and inclusion. An unsafe interpretation is “Applications open Tuesday.” The source instead requires this relationship: Monday is start date. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Before, by, through, until and within define intervals or constraints whose endpoints may differ. The governing source must resolve precise policy boundaries. In this example the central risk is that eligible actions can be rejected, forbidden actions permitted or users notified on the wrong date. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Applications open Monday” discusses. The RELATION and STATUS must support “Monday is start date” while refusing the unsupported alternative “Applications open Tuesday.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Test Monday application. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Monday is start date” and does not casually assert “Applications open Tuesday.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 70: Within 48 hours of receipt

Source claim. The expression “Within 48 hours of receipt” is a short example of deadlines, exclusion and inclusion. An unsafe interpretation is “Within 48 calendar days.” The source instead requires this relationship: Elapsed-hour window. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Before, by, through, until and within define intervals or constraints whose endpoints may differ. The governing source must resolve precise policy boundaries. In this example the central risk is that eligible actions can be rejected, forbidden actions permitted or users notified on the wrong date. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Within 48 hours of receipt” discusses. The RELATION and STATUS must support “Elapsed-hour window” while refusing the unsupported alternative “Within 48 calendar days.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Compute from receipt. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Elapsed-hour window” and does not casually assert “Within 48 calendar days.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 71: Cancel three days ahead

Source claim. The expression “Cancel three days ahead” is a short example of deadlines, exclusion and inclusion. An unsafe interpretation is “Cancel three days afterward.” The source instead requires this relationship: Advance notice precedes event. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Before, by, through, until and within define intervals or constraints whose endpoints may differ. The governing source must resolve precise policy boundaries. In this example the central risk is that eligible actions can be rejected, forbidden actions permitted or users notified on the wrong date. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Cancel three days ahead” discusses. The RELATION and STATUS must support “Advance notice precedes event” while refusing the unsupported alternative “Cancel three days afterward.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Test two-day notice. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Advance notice precedes event” and does not casually assert “Cancel three days afterward.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 72: Embargo expires at midnight

Source claim. The expression “Embargo expires at midnight” is a short example of deadlines, exclusion and inclusion. An unsafe interpretation is “Any nearby midnight is sufficient.” The source instead requires this relationship: Boundary requires exact day and zone. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Before, by, through, until and within define intervals or constraints whose endpoints may differ. The governing source must resolve precise policy boundaries. In this example the central risk is that eligible actions can be rejected, forbidden actions permitted or users notified on the wrong date. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Embargo expires at midnight” discusses. The RELATION and STATUS must support “Boundary requires exact day and zone” while refusing the unsupported alternative “Any nearby midnight is sufficient.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask issuer for timestamp. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Boundary requires exact day and zone” and does not casually assert “Any nearby midnight is sufficient.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

10. Durations, vague time and bounds

Calendar intervals, elapsed hours and approximate periods must not be confused with clock instants or with exact measured durations. A typical failure here is that service expectations or eligibility thresholds change through a small unit or bound error. When analyzing any passage in this family, start with its source evidence, draw the minimal event graph, and decide which potential target-language interpretation must be ruled out. Independent reader testing matters because bilingual reviewers may unconsciously fill missing temporal information from a source text not visible to the final reader.

Case 73: Roughly two hours

Source claim. The expression “Roughly two hours” is a short example of durations, vague time and bounds. An unsafe interpretation is “Exactly two hours.” The source instead requires this relationship: Approximation is explicit. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Calendar intervals, elapsed hours and approximate periods must not be confused with clock instants or with exact measured durations. In this example the central risk is that service expectations or eligibility thresholds change through a small unit or bound error. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Roughly two hours” discusses. The RELATION and STATUS must support “Approximation is explicit” while refusing the unsupported alternative “Exactly two hours.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask for measured maximum. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Approximation is explicit” and does not casually assert “Exactly two hours.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 74: Three business days

Source claim. The expression “Three business days” is a short example of durations, vague time and bounds. An unsafe interpretation is “Exactly 72 hours.” The source instead requires this relationship: Nonworking days affect count. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Calendar intervals, elapsed hours and approximate periods must not be confused with clock instants or with exact measured durations. In this example the central risk is that service expectations or eligibility thresholds change through a small unit or bound error. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Three business days” discusses. The RELATION and STATUS must support “Nonworking days affect count” while refusing the unsupported alternative “Exactly 72 hours.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Test weekend. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Nonworking days affect count” and does not casually assert “Exactly 72 hours.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 75: Motor ran 90 minutes

Source claim. The expression “Motor ran 90 minutes” is a short example of durations, vague time and bounds. An unsafe interpretation is “Motor ran until 01:30.” The source instead requires this relationship: Duration is not clock time. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Calendar intervals, elapsed hours and approximate periods must not be confused with clock instants or with exact measured durations. In this example the central risk is that service expectations or eligibility thresholds change through a small unit or bound error. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Motor ran 90 minutes” discusses. The RELATION and STATUS must support “Duration is not clock time” while refusing the unsupported alternative “Motor ran until 01:30.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask for start. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Duration is not clock time” and does not casually assert “Motor ran until 01:30.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 76: Event lasted into next morning

Source claim. The expression “Event lasted into next morning” is a short example of durations, vague time and bounds. An unsafe interpretation is “Ended at midnight.” The source instead requires this relationship: Interval crosses date boundary. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Calendar intervals, elapsed hours and approximate periods must not be confused with clock instants or with exact measured durations. In this example the central risk is that service expectations or eligibility thresholds change through a small unit or bound error. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Event lasted into next morning” discusses. The RELATION and STATUS must support “Interval crosses date boundary” while refusing the unsupported alternative “Ended at midnight.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Check following date. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Interval crosses date boundary” and does not casually assert “Ended at midnight.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 77: Up to five days

Source claim. The expression “Up to five days” is a short example of durations, vague time and bounds. An unsafe interpretation is “At least five days.” The source instead requires this relationship: Upper rather than lower bound. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Calendar intervals, elapsed hours and approximate periods must not be confused with clock instants or with exact measured durations. In this example the central risk is that service expectations or eligibility thresholds change through a small unit or bound error. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Up to five days” discusses. The RELATION and STATUS must support “Upper rather than lower bound” while refusing the unsupported alternative “At least five days.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Test two-day case. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Upper rather than lower bound” and does not casually assert “At least five days.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 78: More than four hours

Source claim. The expression “More than four hours” is a short example of durations, vague time and bounds. An unsafe interpretation is “No more than four hours.” The source instead requires this relationship: Strict lower bound. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Calendar intervals, elapsed hours and approximate periods must not be confused with clock instants or with exact measured durations. In this example the central risk is that service expectations or eligibility thresholds change through a small unit or bound error. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “More than four hours” discusses. The RELATION and STATUS must support “Strict lower bound” while refusing the unsupported alternative “No more than four hours.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Test five-hour case. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Strict lower bound” and does not casually assert “No more than four hours.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 79: Worked intermittently for weeks

Source claim. The expression “Worked intermittently for weeks” is a short example of durations, vague time and bounds. An unsafe interpretation is “Worked nonstop for weeks.” The source instead requires this relationship: Repeated with gaps. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Calendar intervals, elapsed hours and approximate periods must not be confused with clock instants or with exact measured durations. In this example the central risk is that service expectations or eligibility thresholds change through a small unit or bound error. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Worked intermittently for weeks” discusses. The RELATION and STATUS must support “Repeated with gaps” while refusing the unsupported alternative “Worked nonstop for weeks.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask about pauses. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Repeated with gaps” and does not casually assert “Worked nonstop for weeks.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 80: Should appear within minutes

Source claim. The expression “Should appear within minutes” is a short example of durations, vague time and bounds. An unsafe interpretation is “Guaranteed in two minutes.” The source instead requires this relationship: Approximate expectation. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Calendar intervals, elapsed hours and approximate periods must not be confused with clock instants or with exact measured durations. In this example the central risk is that service expectations or eligibility thresholds change through a small unit or bound error. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Should appear within minutes” discusses. The RELATION and STATUS must support “Approximate expectation” while refusing the unsupported alternative “Guaranteed in two minutes.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask for SLA. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Approximate expectation” and does not casually assert “Guaranteed in two minutes.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

11. Relative dates and calendars

Tomorrow, last week, end of month, working day and fiscal quarter depend on explicit temporal anchors and sometimes domain-specific calendars. A typical failure here is that a relative instruction shifts when the translation is later read or published. When analyzing any passage in this family, start with its source evidence, draw the minimal event graph, and decide which potential target-language interpretation must be ruled out. Independent reader testing matters because bilingual reviewers may unconsciously fill missing temporal information from a source text not visible to the final reader.

Case 81: Tomorrow morning

Source claim. The expression “Tomorrow morning” is a short example of relative dates and calendars. An unsafe interpretation is “Morning after translation publication.” The source instead requires this relationship: Tomorrow anchored to source. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Tomorrow, last week, end of month, working day and fiscal quarter depend on explicit temporal anchors and sometimes domain-specific calendars. In this example the central risk is that a relative instruction shifts when the translation is later read or published. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Tomorrow morning” discusses. The RELATION and STATUS must support “Tomorrow anchored to source” while refusing the unsupported alternative “Morning after translation publication.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Identify source date. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Tomorrow anchored to source” and does not casually assert “Morning after translation publication.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 82: Second week of month

Source claim. The expression “Second week of month” is a short example of relative dates and calendars. An unsafe interpretation is “Week begins on day two.” The source instead requires this relationship: Week number requires convention. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Tomorrow, last week, end of month, working day and fiscal quarter depend on explicit temporal anchors and sometimes domain-specific calendars. In this example the central risk is that a relative instruction shifts when the translation is later read or published. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Second week of month” discusses. The RELATION and STATUS must support “Week number requires convention” while refusing the unsupported alternative “Week begins on day two.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Check week start. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Week number requires convention” and does not casually assert “Week begins on day two.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 83: At quarter end

Source claim. The expression “At quarter end” is a short example of relative dates and calendars. An unsafe interpretation is “At next calendar month end.” The source instead requires this relationship: Quarter may be fiscal. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Tomorrow, last week, end of month, working day and fiscal quarter depend on explicit temporal anchors and sometimes domain-specific calendars. In this example the central risk is that a relative instruction shifts when the translation is later read or published. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “At quarter end” discusses. The RELATION and STATUS must support “Quarter may be fiscal” while refusing the unsupported alternative “At next calendar month end.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask governing calendar. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Quarter may be fiscal” and does not casually assert “At next calendar month end.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 84: Next working day

Source claim. The expression “Next working day” is a short example of relative dates and calendars. An unsafe interpretation is “Exactly twenty-four hours later.” The source instead requires this relationship: Working days skip excluded dates. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Tomorrow, last week, end of month, working day and fiscal quarter depend on explicit temporal anchors and sometimes domain-specific calendars. In this example the central risk is that a relative instruction shifts when the translation is later read or published. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Next working day” discusses. The RELATION and STATUS must support “Working days skip excluded dates” while refusing the unsupported alternative “Exactly twenty-four hours later.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Test weekend. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Working days skip excluded dates” and does not casually assert “Exactly twenty-four hours later.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 85: Just after midnight

Source claim. The expression “Just after midnight” is a short example of relative dates and calendars. An unsafe interpretation is “Just before midnight.” The source instead requires this relationship: Midnight rollover changes date. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Tomorrow, last week, end of month, working day and fiscal quarter depend on explicit temporal anchors and sometimes domain-specific calendars. In this example the central risk is that a relative instruction shifts when the translation is later read or published. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Just after midnight” discusses. The RELATION and STATUS must support “Midnight rollover changes date” while refusing the unsupported alternative “Just before midnight.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Check written date. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Midnight rollover changes date” and does not casually assert “Just before midnight.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 86: In two weeks

Source claim. The expression “In two weeks” is a short example of relative dates and calendars. An unsafe interpretation is “At start of next week.” The source instead requires this relationship: Two-week interval not week start. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Tomorrow, last week, end of month, working day and fiscal quarter depend on explicit temporal anchors and sometimes domain-specific calendars. In this example the central risk is that a relative instruction shifts when the translation is later read or published. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “In two weeks” discusses. The RELATION and STATUS must support “Two-week interval not week start” while refusing the unsupported alternative “At start of next week.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Check reference time. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Two-week interval not week start” and does not casually assert “At start of next week.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 87: Last day of month

Source claim. The expression “Last day of month” is a short example of relative dates and calendars. An unsafe interpretation is “Thirtieth of all months.” The source instead requires this relationship: Month lengths vary. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Tomorrow, last week, end of month, working day and fiscal quarter depend on explicit temporal anchors and sometimes domain-specific calendars. In this example the central risk is that a relative instruction shifts when the translation is later read or published. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Last day of month” discusses. The RELATION and STATUS must support “Month lengths vary” while refusing the unsupported alternative “Thirtieth of all months.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Test February. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Month lengths vary” and does not casually assert “Thirtieth of all months.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 88: Date without year

Source claim. The expression “Date without year” is a short example of relative dates and calendars. An unsafe interpretation is “Automatically this year.” The source instead requires this relationship: Year remains unspecified. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Tomorrow, last week, end of month, working day and fiscal quarter depend on explicit temporal anchors and sometimes domain-specific calendars. In this example the central risk is that a relative instruction shifts when the translation is later read or published. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Date without year” discusses. The RELATION and STATUS must support “Year remains unspecified” while refusing the unsupported alternative “Automatically this year.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask source owner. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Year remains unspecified” and does not casually assert “Automatically this year.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

12. Time zones and civil-time transitions

UTC offsets, named time zones and local wall-clock scheduling are different data types. A named zone can have seasonal or historical rules and ambiguous or nonexistent clock readings. A typical failure here is that the same appointment is silently shifted by an hour or onto a different calendar day. When analyzing any passage in this family, start with its source evidence, draw the minimal event graph, and decide which potential target-language interpretation must be ruled out. Independent reader testing matters because bilingual reviewers may unconsciously fill missing temporal information from a source text not visible to the final reader.

Case 89: Meeting at 16:00 UTC

Source claim. The expression “Meeting at 16:00 UTC” is a short example of time zones and civil-time transitions. An unsafe interpretation is “16:00 in every zone.” The source instead requires this relationship: UTC needs local conversion. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. UTC offsets, named time zones and local wall-clock scheduling are different data types. A named zone can have seasonal or historical rules and ambiguous or nonexistent clock readings. In this example the central risk is that the same appointment is silently shifted by an hour or onto a different calendar day. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Meeting at 16:00 UTC” discusses. The RELATION and STATUS must support “UTC needs local conversion” while refusing the unsupported alternative “16:00 in every zone.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Check two zone displays. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “UTC needs local conversion” and does not casually assert “16:00 in every zone.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 90: 09:00 in organizer zone

Source claim. The expression “09:00 in organizer zone” is a short example of time zones and civil-time transitions. An unsafe interpretation is “09:00 in attendee zone.” The source instead requires this relationship: Time belongs to specified civil zone. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. UTC offsets, named time zones and local wall-clock scheduling are different data types. A named zone can have seasonal or historical rules and ambiguous or nonexistent clock readings. In this example the central risk is that the same appointment is silently shifted by an hour or onto a different calendar day. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “09:00 in organizer zone” discusses. The RELATION and STATUS must support “Time belongs to specified civil zone” while refusing the unsupported alternative “09:00 in attendee zone.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask organizer zone. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Time belongs to specified civil zone” and does not casually assert “09:00 in attendee zone.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 91: Timestamp offset +02:00

Source claim. The expression “Timestamp offset +02:00” is a short example of time zones and civil-time transitions. An unsafe interpretation is “Zone always UTC+02.” The source instead requires this relationship: Offset only at that instant. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. UTC offsets, named time zones and local wall-clock scheduling are different data types. A named zone can have seasonal or historical rules and ambiguous or nonexistent clock readings. In this example the central risk is that the same appointment is silently shifted by an hour or onto a different calendar day. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Timestamp offset +02:00” discusses. The RELATION and STATUS must support “Offset only at that instant” while refusing the unsupported alternative “Zone always UTC+02.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Test another season. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Offset only at that instant” and does not casually assert “Zone always UTC+02.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 92: Autumn clock overlap

Source claim. The expression “Autumn clock overlap” is a short example of time zones and civil-time transitions. An unsafe interpretation is “One unique instant guaranteed.” The source instead requires this relationship: Repeated wall time may need offset. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. UTC offsets, named time zones and local wall-clock scheduling are different data types. A named zone can have seasonal or historical rules and ambiguous or nonexistent clock readings. In this example the central risk is that the same appointment is silently shifted by an hour or onto a different calendar day. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Autumn clock overlap” discusses. The RELATION and STATUS must support “Repeated wall time may need offset” while refusing the unsupported alternative “One unique instant guaranteed.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask for disambiguation. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Repeated wall time may need offset” and does not casually assert “One unique instant guaranteed.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 93: Spring clock gap

Source claim. The expression “Spring clock gap” is a short example of time zones and civil-time transitions. An unsafe interpretation is “Every wall time is possible.” The source instead requires this relationship: Some local times do not exist. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. UTC offsets, named time zones and local wall-clock scheduling are different data types. A named zone can have seasonal or historical rules and ambiguous or nonexistent clock readings. In this example the central risk is that the same appointment is silently shifted by an hour or onto a different calendar day. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Spring clock gap” discusses. The RELATION and STATUS must support “Some local times do not exist” while refusing the unsupported alternative “Every wall time is possible.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Check transition policy. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Some local times do not exist” and does not casually assert “Every wall time is possible.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 94: Late-night timestamp Z

Source claim. The expression “Late-night timestamp Z” is a short example of time zones and civil-time transitions. An unsafe interpretation is “Same calendar date everywhere.” The source instead requires this relationship: Local date can change. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. UTC offsets, named time zones and local wall-clock scheduling are different data types. A named zone can have seasonal or historical rules and ambiguous or nonexistent clock readings. In this example the central risk is that the same appointment is silently shifted by an hour or onto a different calendar day. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Late-night timestamp Z” discusses. The RELATION and STATUS must support “Local date can change” while refusing the unsupported alternative “Same calendar date everywhere.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Compare display dates. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Local date can change” and does not casually assert “Same calendar date everywhere.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 95: Numeric +00:00

Source claim. The expression “Numeric +00:00” is a short example of time zones and civil-time transitions. An unsafe interpretation is “Specific named zone implied.” The source instead requires this relationship: Offset does not identify zone. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. UTC offsets, named time zones and local wall-clock scheduling are different data types. A named zone can have seasonal or historical rules and ambiguous or nonexistent clock readings. In this example the central risk is that the same appointment is silently shifted by an hour or onto a different calendar day. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Numeric +00:00” discusses. The RELATION and STATUS must support “Offset does not identify zone” while refusing the unsupported alternative “Specific named zone implied.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Check metadata. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Offset does not identify zone” and does not casually assert “Specific named zone implied.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 96: Weekly 09:00 local

Source claim. The expression “Weekly 09:00 local” is a short example of time zones and civil-time transitions. An unsafe interpretation is “Weekly fixed UTC hour.” The source instead requires this relationship: Civil recurrence can shift UTC hour. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. UTC offsets, named time zones and local wall-clock scheduling are different data types. A named zone can have seasonal or historical rules and ambiguous or nonexistent clock readings. In this example the central risk is that the same appointment is silently shifted by an hour or onto a different calendar day. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Weekly 09:00 local” discusses. The RELATION and STATUS must support “Civil recurrence can shift UTC hour” while refusing the unsupported alternative “Weekly fixed UTC hour.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Check seasonal meetings. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Civil recurrence can shift UTC hour” and does not casually assert “Weekly fixed UTC hour.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

13. Localized dates and runtime language

Locale patterns and runtime placeholders are not ordinary strings. Unicode CLDR supports display conventions, but the complete rendered phrase must still communicate the same time and grammatical relation. A typical failure here is that correct stored values become ambiguous, missing or misleading in the user interface. When analyzing any passage in this family, start with its source evidence, draw the minimal event graph, and decide which potential target-language interpretation must be ruled out. Independent reader testing matters because bilingual reviewers may unconsciously fill missing temporal information from a source text not visible to the final reader.

Case 97: 03/04/2026 without locale

Source claim. The expression “03/04/2026 without locale” is a short example of localized dates and runtime language. An unsafe interpretation is “Universal day-month interpretation.” The source instead requires this relationship: Numeric order is ambiguous. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Locale patterns and runtime placeholders are not ordinary strings. Unicode CLDR supports display conventions, but the complete rendered phrase must still communicate the same time and grammatical relation. In this example the central risk is that correct stored values become ambiguous, missing or misleading in the user interface. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “03/04/2026 without locale” discusses. The RELATION and STATUS must support “Numeric order is ambiguous” while refusing the unsupported alternative “Universal day-month interpretation.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Use month name. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Numeric order is ambiguous” and does not casually assert “Universal day-month interpretation.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 98: {date} at {time}

Source claim. The expression “{date} at {time}” is a short example of localized dates and runtime language. An unsafe interpretation is “Delete placeholders.” The source instead requires this relationship: Tokens must survive translation. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Locale patterns and runtime placeholders are not ordinary strings. Unicode CLDR supports display conventions, but the complete rendered phrase must still communicate the same time and grammatical relation. In this example the central risk is that correct stored values become ambiguous, missing or misleading in the user interface. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “{date} at {time}” discusses. The RELATION and STATUS must support “Tokens must survive translation” while refusing the unsupported alternative “Delete placeholders.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Render actual values. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Tokens must survive translation” and does not casually assert “Delete placeholders.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 99: Tomorrow at {time}

Source claim. The expression “Tomorrow at {time}” is a short example of localized dates and runtime language. An unsafe interpretation is “Use isolated date grammar.” The source instead requires this relationship: Relative-date combination may differ. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Locale patterns and runtime placeholders are not ordinary strings. Unicode CLDR supports display conventions, but the complete rendered phrase must still communicate the same time and grammatical relation. In this example the central risk is that correct stored values become ambiguous, missing or misleading in the user interface. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Tomorrow at {time}” discusses. The RELATION and STATUS must support “Relative-date combination may differ” while refusing the unsupported alternative “Use isolated date grammar.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Check full phrase. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Relative-date combination may differ” and does not casually assert “Use isolated date grammar.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 100: 09:00–11:00 interval

Source claim. The expression “09:00–11:00 interval” is a short example of localized dates and runtime language. An unsafe interpretation is “Only 09:00 exists.” The source instead requires this relationship: Both endpoints define range. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Locale patterns and runtime placeholders are not ordinary strings. Unicode CLDR supports display conventions, but the complete rendered phrase must still communicate the same time and grammatical relation. In this example the central risk is that correct stored values become ambiguous, missing or misleading in the user interface. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “09:00–11:00 interval” discusses. The RELATION and STATUS must support “Both endpoints define range” while refusing the unsupported alternative “Only 09:00 exists.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Check start and end. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Both endpoints define range” and does not casually assert “Only 09:00 exists.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 101: 12-hour display with AM

Source claim. The expression “12-hour display with AM” is a short example of localized dates and runtime language. An unsafe interpretation is “AM can disappear safely.” The source instead requires this relationship: Marker disambiguates. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Locale patterns and runtime placeholders are not ordinary strings. Unicode CLDR supports display conventions, but the complete rendered phrase must still communicate the same time and grammatical relation. In this example the central risk is that correct stored values become ambiguous, missing or misleading in the user interface. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “12-hour display with AM” discusses. The RELATION and STATUS must support “Marker disambiguates” while refusing the unsupported alternative “AM can disappear safely.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Test noon and midnight. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Marker disambiguates” and does not casually assert “AM can disappear safely.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 102: Month name in sentence

Source claim. The expression “Month name in sentence” is a short example of localized dates and runtime language. An unsafe interpretation is “Always use standalone month form.” The source instead requires this relationship: Grammatical inflection may vary. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Locale patterns and runtime placeholders are not ordinary strings. Unicode CLDR supports display conventions, but the complete rendered phrase must still communicate the same time and grammatical relation. In this example the central risk is that correct stored values become ambiguous, missing or misleading in the user interface. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Month name in sentence” discusses. The RELATION and STATUS must support “Grammatical inflection may vary” while refusing the unsupported alternative “Always use standalone month form.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Read complete sentence. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Grammatical inflection may vary” and does not casually assert “Always use standalone month form.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 103: Updated two minutes ago

Source claim. The expression “Updated two minutes ago” is a short example of localized dates and runtime language. An unsafe interpretation is “Cache wording forever.” The source instead requires this relationship: Relative display changes with clock. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Locale patterns and runtime placeholders are not ordinary strings. Unicode CLDR supports display conventions, but the complete rendered phrase must still communicate the same time and grammatical relation. In this example the central risk is that correct stored values become ambiguous, missing or misleading in the user interface. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Updated two minutes ago” discusses. The RELATION and STATUS must support “Relative display changes with clock” while refusing the unsupported alternative “Cache wording forever.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Retest later. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Relative display changes with clock” and does not casually assert “Cache wording forever.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 104: Spoken date label

Source claim. The expression “Spoken date label” is a short example of localized dates and runtime language. An unsafe interpretation is “Only visual digit string matters.” The source instead requires this relationship: Accessible rendering must be clear. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. Locale patterns and runtime placeholders are not ordinary strings. Unicode CLDR supports display conventions, but the complete rendered phrase must still communicate the same time and grammatical relation. In this example the central risk is that correct stored values become ambiguous, missing or misleading in the user interface. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Spoken date label” discusses. The RELATION and STATUS must support “Accessible rendering must be clear” while refusing the unsupported alternative “Only visual digit string matters.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Listen via screen reader. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Accessible rendering must be clear” and does not casually assert “Only visual digit string matters.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

14. Version history, archived claims and knowledge

The time a report was issued, when an event happened, when a correction became effective and when evidence became known are different temporal facts. A typical failure here is that the past is retroactively rewritten or new knowledge falsely attributed to an earlier actor. When analyzing any passage in this family, start with its source evidence, draw the minimal event graph, and decide which potential target-language interpretation must be ruled out. Independent reader testing matters because bilingual reviewers may unconsciously fill missing temporal information from a source text not visible to the final reader.

Case 105: Policy issued Monday effective Friday

Source claim. The expression “Policy issued Monday effective Friday” is a short example of version history, archived claims and knowledge. An unsafe interpretation is “Policy active Monday.” The source instead requires this relationship: Issue and effective dates differ. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. The time a report was issued, when an event happened, when a correction became effective and when evidence became known are different temporal facts. In this example the central risk is that the past is retroactively rewritten or new knowledge falsely attributed to an earlier actor. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Policy issued Monday effective Friday” discusses. The RELATION and STATUS must support “Issue and effective dates differ” while refusing the unsupported alternative “Policy active Monday.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Check obligations start. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Issue and effective dates differ” and does not casually assert “Policy active Monday.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 106: Observed 10:00 filed 14:00

Source claim. The expression “Observed 10:00 filed 14:00” is a short example of version history, archived claims and knowledge. An unsafe interpretation is “Observed at 14:00.” The source instead requires this relationship: Event and entry times differ. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. The time a report was issued, when an event happened, when a correction became effective and when evidence became known are different temporal facts. In this example the central risk is that the past is retroactively rewritten or new knowledge falsely attributed to an earlier actor. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Observed 10:00 filed 14:00” discusses. The RELATION and STATUS must support “Event and entry times differ” while refusing the unsupported alternative “Observed at 14:00.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask event timestamp. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Event and entry times differ” and does not casually assert “Observed at 14:00.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 107: Later correction published

Source claim. The expression “Later correction published” is a short example of version history, archived claims and knowledge. An unsafe interpretation is “Correction part of first version.” The source instead requires this relationship: Revision occurred afterward. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. The time a report was issued, when an event happened, when a correction became effective and when evidence became known are different temporal facts. In this example the central risk is that the past is retroactively rewritten or new knowledge falsely attributed to an earlier actor. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Later correction published” discusses. The RELATION and STATUS must support “Revision occurred afterward” while refusing the unsupported alternative “Correction part of first version.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Compare archives. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Revision occurred afterward” and does not casually assert “Correction part of first version.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 108: Old version expired April

Source claim. The expression “Old version expired April” is a short example of version history, archived claims and knowledge. An unsafe interpretation is “All versions expired April.” The source instead requires this relationship: Validity is version-specific. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. The time a report was issued, when an event happened, when a correction became effective and when evidence became known are different temporal facts. In this example the central risk is that the past is retroactively rewritten or new knowledge falsely attributed to an earlier actor. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Old version expired April” discusses. The RELATION and STATUS must support “Validity is version-specific” while refusing the unsupported alternative “All versions expired April.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask which version. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Validity is version-specific” and does not casually assert “All versions expired April.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 109: Page retrieved yesterday

Source claim. The expression “Page retrieved yesterday” is a short example of version history, archived claims and knowledge. An unsafe interpretation is “Event occurred yesterday.” The source instead requires this relationship: Access time isn’t event time. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. The time a report was issued, when an event happened, when a correction became effective and when evidence became known are different temporal facts. In this example the central risk is that the past is retroactively rewritten or new knowledge falsely attributed to an earlier actor. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Page retrieved yesterday” discusses. The RELATION and STATUS must support “Access time isn’t event time” while refusing the unsupported alternative “Event occurred yesterday.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask actual event date. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Access time isn’t event time” and does not casually assert “Event occurred yesterday.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 110: Investigators did not know yet

Source claim. The expression “Investigators did not know yet” is a short example of version history, archived claims and knowledge. An unsafe interpretation is “Investigators knew from start.” The source instead requires this relationship: Later discovery cannot be backdated. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. The time a report was issued, when an event happened, when a correction became effective and when evidence became known are different temporal facts. In this example the central risk is that the past is retroactively rewritten or new knowledge falsely attributed to an earlier actor. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Investigators did not know yet” discusses. The RELATION and STATUS must support “Later discovery cannot be backdated” while refusing the unsupported alternative “Investigators knew from start.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Check available evidence. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Later discovery cannot be backdated” and does not casually assert “Investigators knew from start.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 111: Early estimate later revised

Source claim. The expression “Early estimate later revised” is a short example of version history, archived claims and knowledge. An unsafe interpretation is “Original estimate already precise.” The source instead requires this relationship: Knowledge changed over time. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. The time a report was issued, when an event happened, when a correction became effective and when evidence became known are different temporal facts. In this example the central risk is that the past is retroactively rewritten or new knowledge falsely attributed to an earlier actor. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Early estimate later revised” discusses. The RELATION and STATUS must support “Knowledge changed over time” while refusing the unsupported alternative “Original estimate already precise.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Compare estimates. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Knowledge changed over time” and does not casually assert “Original estimate already precise.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 112: Snapshot taken month end

Source claim. The expression “Snapshot taken month end” is a short example of version history, archived claims and knowledge. An unsafe interpretation is “Includes all later submissions.” The source instead requires this relationship: Cutoff constrains observations. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. The time a report was issued, when an event happened, when a correction became effective and when evidence became known are different temporal facts. In this example the central risk is that the past is retroactively rewritten or new knowledge falsely attributed to an earlier actor. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Snapshot taken month end” discusses. The RELATION and STATUS must support “Cutoff constrains observations” while refusing the unsupported alternative “Includes all later submissions.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Test late submission. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Cutoff constrains observations” and does not casually assert “Includes all later submissions.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

15. Narrative sequence and reader decisions

A story can present events out of order, while an operational passage may require a particular action sequence. In both cases, the source-supported relation outranks sentence order. A typical failure here is that a reader reconstructs an invented timeline or acts before the required condition. When analyzing any passage in this family, start with its source evidence, draw the minimal event graph, and decide which potential target-language interpretation must be ruled out. Independent reader testing matters because bilingual reviewers may unconsciously fill missing temporal information from a source text not visible to the final reader.

Case 113: Story opens in aftermath then returns earlier

Source claim. The expression “Story opens in aftermath then returns earlier” is a short example of narrative sequence and reader decisions. An unsafe interpretation is “Events occur in text order.” The source instead requires this relationship: Narration is nonlinear. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. A story can present events out of order, while an operational passage may require a particular action sequence. In both cases, the source-supported relation outranks sentence order. In this example the central risk is that a reader reconstructs an invented timeline or acts before the required condition. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Story opens in aftermath then returns earlier” discusses. The RELATION and STATUS must support “Narration is nonlinear” while refusing the unsupported alternative “Events occur in text order.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Draw story chronology. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Narration is nonlinear” and does not casually assert “Events occur in text order.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 114: She remembered room from years earlier

Source claim. The expression “She remembered room from years earlier” is a short example of narrative sequence and reader decisions. An unsafe interpretation is “Room looks that way now.” The source instead requires this relationship: Recollection refers to past. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. A story can present events out of order, while an operational passage may require a particular action sequence. In both cases, the source-supported relation outranks sentence order. In this example the central risk is that a reader reconstructs an invented timeline or acts before the required condition. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “She remembered room from years earlier” discusses. The RELATION and STATUS must support “Recollection refers to past” while refusing the unsupported alternative “Room looks that way now.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask present appearance. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Recollection refers to past” and does not casually assert “Room looks that way now.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 115: He would later regret the remark

Source claim. The expression “He would later regret the remark” is a short example of narrative sequence and reader decisions. An unsafe interpretation is “He regretted remark while saying it.” The source instead requires this relationship: Regret is later than speech. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. A story can present events out of order, while an operational passage may require a particular action sequence. In both cases, the source-supported relation outranks sentence order. In this example the central risk is that a reader reconstructs an invented timeline or acts before the required condition. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “He would later regret the remark” discusses. The RELATION and STATUS must support “Regret is later than speech” while refusing the unsupported alternative “He regretted remark while saying it.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask regret onset. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Regret is later than speech” and does not casually assert “He regretted remark while saying it.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 116: Meanwhile other team waited

Source claim. The expression “Meanwhile other team waited” is a short example of narrative sequence and reader decisions. An unsafe interpretation is “Other team waited afterward.” The source instead requires this relationship: Scenes overlap. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. A story can present events out of order, while an operational passage may require a particular action sequence. In both cases, the source-supported relation outranks sentence order. In this example the central risk is that a reader reconstructs an invented timeline or acts before the required condition. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Meanwhile other team waited” discusses. The RELATION and STATUS must support “Scenes overlap” while refusing the unsupported alternative “Other team waited afterward.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Compare scene intervals. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Scenes overlap” and does not casually assert “Other team waited afterward.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 117: Flashback ends at opening scene

Source claim. The expression “Flashback ends at opening scene” is a short example of narrative sequence and reader decisions. An unsafe interpretation is “Flashback continues beyond ending.” The source instead requires this relationship: Return point is earlier scene. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. A story can present events out of order, while an operational passage may require a particular action sequence. In both cases, the source-supported relation outranks sentence order. In this example the central risk is that a reader reconstructs an invented timeline or acts before the required condition. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Flashback ends at opening scene” discusses. The RELATION and STATUS must support “Return point is earlier scene” while refusing the unsupported alternative “Flashback continues beyond ending.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Mark narrative boundary. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Return point is earlier scene” and does not casually assert “Flashback continues beyond ending.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 118: Appointment provisional until accepted

Source claim. The expression “Appointment provisional until accepted” is a short example of narrative sequence and reader decisions. An unsafe interpretation is “Appointment confirmed when requested.” The source instead requires this relationship: Acceptance gate controls status. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. A story can present events out of order, while an operational passage may require a particular action sequence. In both cases, the source-supported relation outranks sentence order. In this example the central risk is that a reader reconstructs an invented timeline or acts before the required condition. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Appointment provisional until accepted” discusses. The RELATION and STATUS must support “Acceptance gate controls status” while refusing the unsupported alternative “Appointment confirmed when requested.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Check pending booking. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Acceptance gate controls status” and does not casually assert “Appointment confirmed when requested.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 119: Shipment after inspection passes

Source claim. The expression “Shipment after inspection passes” is a short example of narrative sequence and reader decisions. An unsafe interpretation is “Shipment while inspection underway.” The source instead requires this relationship: Pass precedes permission. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. A story can present events out of order, while an operational passage may require a particular action sequence. In both cases, the source-supported relation outranks sentence order. In this example the central risk is that a reader reconstructs an invented timeline or acts before the required condition. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Shipment after inspection passes” discusses. The RELATION and STATUS must support “Pass precedes permission” while refusing the unsupported alternative “Shipment while inspection underway.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Test failed inspection. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Pass precedes permission” and does not casually assert “Shipment while inspection underway.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Case 120: Alert supersedes previous notice

Source claim. The expression “Alert supersedes previous notice” is a short example of narrative sequence and reader decisions. An unsafe interpretation is “Both notices always active.” The source instead requires this relationship: Later advice replaces earlier. The translator should state that relationship in a private review note before choosing wording. Two constructions can be equally idiomatic yet not equally accurate, because a target reader can infer a different order, completion state, frequency, boundary or level of evidence.

Hidden mechanism. A story can present events out of order, while an operational passage may require a particular action sequence. In both cases, the source-supported relation outranks sentence order. In this example the central risk is that a reader reconstructs an invented timeline or acts before the required condition. Examine whether the source actually asserts a dated event, a state at a particular vantage point, a continuing process, a recurring pattern or a future possibility. Mark explicit temporal words but also consult surrounding material that supplies a speaker date, procedural prerequisite or earlier reference event. Do not replace an unknown fact with a confident timestamp, completion claim or causal theory simply to make the translated sentence read more smoothly.

Repeatable repair. Create a record with columns EVENT, REFERENCE, RELATION, STATUS, PRECISION and EVIDENCE. The EVENT is what “Alert supersedes previous notice” discusses. The RELATION and STATUS must support “Later advice replaces earlier” while refusing the unsupported alternative “Both notices always active.” Draft the target sentence using whatever time expressions, particles, auxiliaries or clause order that language naturally provides. If a preceding paragraph alters the reference point, carry that evidence through the translation unit. Retain source uncertainty and use an explicit clarification request when a material boundary cannot be determined.

Worked reader diagnostic. Ask which applies. Give the translated sentence to a qualified independent reader who has not been shown the source, and ask for the sequence of events or practical action that follows. A pass means the answer is compatible with “Later advice replaces earlier” and does not casually assert “Both notices always active.” If the unsafe reading emerges, log a semantic time error rather than merely an awkward sentence. Revise the phrasing and retest at least one related passage to ensure consistent treatment of the same reference, recurrence or deadline across the document.

Transfer test. Replace the key time expression with a contrary form—before for after, finished for underway, never for not yet, or confirmed for expected—and examine how the reader’s answer changes. A robust translation method should preserve the original contrast, not produce one vague output for both meanings. Store the supporting source sentence, the reviewed target expression, its reference anchor and the reader’s paraphrase. This evidence helps another reviewer detect recurrence of the same error after a machine-translation update, source revision or editorial rewrite.

Applied laboratories: from translation theory to release decisions

Laboratory 1: Incident chronology

Initial observation. A log records an inspection, a warning and a report at separate times. An earlier event is mentioned in a later sentence. Establish what the text says and what it does not say; do not allow an automatic model, a reviewer or a user-interface default to decide the missing facts without evidence. The central goal is a target passage that supports the same justified action or understanding.

Worked method. Make a table for event time, observation time and report time. Keep BEFORE and AFTER edges separate from guesses about causes. Build a temporal evidence sheet with source span, event identity, reference time, asserted edges, certainty, deadline or zone metadata and a target reader question. Distinguish what is translation from what is software formatting or an authorized policy clarification.

Diagnostic and release decision. Ask which event occurred first and what was known at the inspection. Reject a translation that assigns report time to the event. Have a second reviewer answer this without seeing the source. Preserve the answer, version and source evidence. A grammatical rewrite does not pass if it changes status, event order, timeline or the reader’s permission to act. Where consequences are high, obtain qualified domain and source-owner review before publishing.

Laboratory 2: Ongoing compliance review

Initial observation. A report says the team has been auditing for weeks but gives no approval record. Establish what the text says and what it does not say; do not allow an automatic model, a reviewer or a user-interface default to decide the missing facts without evidence. The central goal is a target passage that supports the same justified action or understanding.

Worked method. Keep prolonged activity separate from completed certification. Request sign-off evidence before using approved or finished language. Build a temporal evidence sheet with source span, event identity, reference time, asserted edges, certainty, deadline or zone metadata and a target reader question. Distinguish what is translation from what is software formatting or an authorized policy clarification.

Diagnostic and release decision. Target-only readers must still understand that additional work may remain. Have a second reviewer answer this without seeing the source. Preserve the answer, version and source evidence. A grammatical rewrite does not pass if it changes status, event order, timeline or the reader’s permission to act. Where consequences are high, obtain qualified domain and source-owner review before publishing.

Laboratory 3: Dated quotation

Initial observation. An archived notice says the schedule will be updated tomorrow. Establish what the text says and what it does not say; do not allow an automatic model, a reviewer or a user-interface default to decide the missing facts without evidence. The central goal is a target passage that supports the same justified action or understanding.

Worked method. Anchor tomorrow to the original notice time. Preserve source date or an authorized explicit equivalent. Build a temporal evidence sheet with source span, event identity, reference time, asserted edges, certainty, deadline or zone metadata and a target reader question. Distinguish what is translation from what is software formatting or an authorized policy clarification.

Diagnostic and release decision. The translation must not promise an update tomorrow relative to today’s reader. Have a second reviewer answer this without seeing the source. Preserve the answer, version and source evidence. A grammatical rewrite does not pass if it changes status, event order, timeline or the reader’s permission to act. Where consequences are high, obtain qualified domain and source-owner review before publishing.

Laboratory 4: Cross-zone appointment

Initial observation. A recurring meeting is set at 09:00 in a named civil-time zone. Establish what the text says and what it does not say; do not allow an automatic model, a reviewer or a user-interface default to decide the missing facts without evidence. The central goal is a target passage that supports the same justified action or understanding.

Worked method. Keep local wall time and named zone; generate each recurrence under the zone rules for its date. Build a temporal evidence sheet with source span, event identity, reference time, asserted edges, certainty, deadline or zone metadata and a target reader question. Distinguish what is translation from what is software formatting or an authorized policy clarification.

Diagnostic and release decision. Test before and after a possible offset change, and verify the same local intended hour. Have a second reviewer answer this without seeing the source. Preserve the answer, version and source evidence. A grammatical rewrite does not pass if it changes status, event order, timeline or the reader’s permission to act. Where consequences are high, obtain qualified domain and source-owner review before publishing.

Laboratory 5: Strict deadline

Initial observation. A document says submissions must arrive before an hour but omits the governing time zone. Establish what the text says and what it does not say; do not allow an automatic model, a reviewer or a user-interface default to decide the missing facts without evidence. The central goal is a target passage that supports the same justified action or understanding.

Worked method. Preserve strict earlier-than and escalate missing zone information to the owner. Build a temporal evidence sheet with source span, event identity, reference time, asserted edges, certainty, deadline or zone metadata and a target reader question. Distinguish what is translation from what is software formatting or an authorized policy clarification.

Diagnostic and release decision. Test seconds on either side of the authorized cutoff once defined. Have a second reviewer answer this without seeing the source. Preserve the answer, version and source evidence. A grammatical rewrite does not pass if it changes status, event order, timeline or the reader’s permission to act. Where consequences are high, obtain qualified domain and source-owner review before publishing.

Laboratory 6: Narrative flashback

Initial observation. A story describes an aftermath and then returns to its cause. Establish what the text says and what it does not say; do not allow an automatic model, a reviewer or a user-interface default to decide the missing facts without evidence. The central goal is a target passage that supports the same justified action or understanding.

Worked method. Separate story time from telling order and preserve a natural entrance and exit from retrospection. Build a temporal evidence sheet with source span, event identity, reference time, asserted edges, certainty, deadline or zone metadata and a target reader question. Distinguish what is translation from what is software formatting or an authorized policy clarification.

Diagnostic and release decision. Ask readers which event happened first in the fictional world. Have a second reviewer answer this without seeing the source. Preserve the answer, version and source evidence. A grammatical rewrite does not pass if it changes status, event order, timeline or the reader’s permission to act. Where consequences are high, obtain qualified domain and source-owner review before publishing.

Laboratory 7: Promise versus outcome

Initial observation. A dated plan expected delivery by a quarter end, but the outcome was delayed. Establish what the text says and what it does not say; do not allow an automatic model, a reviewer or a user-interface default to decide the missing facts without evidence. The central goal is a target passage that supports the same justified action or understanding.

Worked method. Keep planned and actual completion as two distinct claims with separate source versions. Build a temporal evidence sheet with source span, event identity, reference time, asserted edges, certainty, deadline or zone metadata and a target reader question. Distinguish what is translation from what is software formatting or an authorized policy clarification.

Diagnostic and release decision. Ask whether the earlier plan proved that delivery occurred; it did not. Have a second reviewer answer this without seeing the source. Preserve the answer, version and source evidence. A grammatical rewrite does not pass if it changes status, event order, timeline or the reader’s permission to act. Where consequences are high, obtain qualified domain and source-owner review before publishing.

Laboratory 8: Civil midnight

Initial observation. An event starts shortly before midnight and continues into the next day. Establish what the text says and what it does not say; do not allow an automatic model, a reviewer or a user-interface default to decide the missing facts without evidence. The central goal is a target passage that supports the same justified action or understanding.

Worked method. Retain date rollover and determine whether the clock belongs to the stated zone. Build a temporal evidence sheet with source span, event identity, reference time, asserted edges, certainty, deadline or zone metadata and a target reader question. Distinguish what is translation from what is software formatting or an authorized policy clarification.

Diagnostic and release decision. Test both calendar dates in the final display. Have a second reviewer answer this without seeing the source. Preserve the answer, version and source evidence. A grammatical rewrite does not pass if it changes status, event order, timeline or the reader’s permission to act. Where consequences are high, obtain qualified domain and source-owner review before publishing.

Laboratory 9: Date-only license

Initial observation. A license is valid through the final day of a month, while its system stores an exclusive end instant. Establish what the text says and what it does not say; do not allow an automatic model, a reviewer or a user-interface default to decide the missing facts without evidence. The central goal is a target passage that supports the same justified action or understanding.

Worked method. Retain the reader’s valid last day and verify the technical instant implements it under the authorized zone. Build a temporal evidence sheet with source span, event identity, reference time, asserted edges, certainty, deadline or zone metadata and a target reader question. Distinguish what is translation from what is software formatting or an authorized policy clarification.

Diagnostic and release decision. Test access near the last day’s endpoint and after the next day’s start. Have a second reviewer answer this without seeing the source. Preserve the answer, version and source evidence. A grammatical rewrite does not pass if it changes status, event order, timeline or the reader’s permission to act. Where consequences are high, obtain qualified domain and source-owner review before publishing.

Laboratory 10: Evidence arrival

Initial observation. An investigator’s later findings are included in a retrospective report. Establish what the text says and what it does not say; do not allow an automatic model, a reviewer or a user-interface default to decide the missing facts without evidence. The central goal is a target passage that supports the same justified action or understanding.

Worked method. Separate the time of the event from when evidence arrived and what the investigator knew earlier. Build a temporal evidence sheet with source span, event identity, reference time, asserted edges, certainty, deadline or zone metadata and a target reader question. Distinguish what is translation from what is software formatting or an authorized policy clarification.

Diagnostic and release decision. Ask what was knowable at the decision time. Do not backdate the evidence. Have a second reviewer answer this without seeing the source. Preserve the answer, version and source evidence. A grammatical rewrite does not pass if it changes status, event order, timeline or the reader’s permission to act. Where consequences are high, obtain qualified domain and source-owner review before publishing.

Laboratory 11: Uncertain order

Initial observation. Conflicting witness reports give different orders and no synchronized timestamps. Establish what the text says and what it does not say; do not allow an automatic model, a reviewer or a user-interface default to decide the missing facts without evidence. The central goal is a target passage that supports the same justified action or understanding.

Worked method. Record the order as contested and translate each attributed account faithfully. Build a temporal evidence sheet with source span, event identity, reference time, asserted edges, certainty, deadline or zone metadata and a target reader question. Distinguish what is translation from what is software formatting or an authorized policy clarification.

Diagnostic and release decision. The target must not invent an agreed chronology. Have a second reviewer answer this without seeing the source. Preserve the answer, version and source evidence. A grammatical rewrite does not pass if it changes status, event order, timeline or the reader’s permission to act. Where consequences are high, obtain qualified domain and source-owner review before publishing.

Laboratory 12: Localized interface

Initial observation. A numeric date appears in a different locale inside a translated sentence. Establish what the text says and what it does not say; do not allow an automatic model, a reviewer or a user-interface default to decide the missing facts without evidence. The central goal is a target passage that supports the same justified action or understanding.

Worked method. Retain a structured value and use appropriate CLDR-like formatting; review the complete rendered sentence and accessible label. Build a temporal evidence sheet with source span, event identity, reference time, asserted edges, certainty, deadline or zone metadata and a target reader question. Distinguish what is translation from what is software formatting or an authorized policy clarification.

Diagnostic and release decision. Have two independent readers identify day, month and year correctly. Have a second reviewer answer this without seeing the source. Preserve the answer, version and source evidence. A grammatical rewrite does not pass if it changes status, event order, timeline or the reader’s permission to act. Where consequences are high, obtain qualified domain and source-owner review before publishing.

The temporal worksheet to copy into any review

For each important segment record SOURCE, VERSION, AUTHOR/QUOTED SPEAKER, SPEECH TIME, REFERENCE TIME, EVENT, EVENT INTERVAL, BEFORE/AFTER/OVERLAP RELATIONS, COMPLETION OR PROGRESSION, RECURRENCE, MODAL CERTAINTY, OBSERVATION TIME, KNOWLEDGE TIME, PRECISION, CALENDAR, UTC INSTANT IF KNOWN, CIVIL TIME ZONE IF KNOWN, DEADLINE INCLUSION, TARGET WORDING, COUNTERFACTUAL READING, READER-TEST RESULT and RELEASE DECISION. Leave genuinely unspecified fields unknown. Never turn an empty cell into a fact because the target grammar seems to require one.

A smaller version works for ordinary prose: EVENT, REFERENCE, RELATION, STATE and CERTAINTY. The larger form is valuable for contracts, historical texts, incident logs, international scheduling and production software. Compare the source and target as independent event graphs. Preserve every source-supported edge, preserve unresolved relations as unresolved, and check that a fluent target does not invite an extra outcome that the source does not state.

Automation and human quality assurance

Software can catch corrupted date tokens, changed numerals, invalid timestamps, missing relative-time variables and inconsistent event annotations. Test midnight, calendar boundaries, recurring civil-time appointments, daylight-saving gaps and overlaps when the application uses those features. Reviewers must still judge meaning: even a valid timestamp may be attached to the wrong event, and fluent machine translation can erase a before–after relationship or convert a historical intention into a current commitment.

Translation memories have similar risks. A unit previously translated as tomorrow may be carried into a new issue date; a verified-complete sentence may be reused while the current source says checking is underway. Save context and event metadata with the translation, re-evaluate changed sources, and ask target-only readers practical questions. In high-consequence medical, legal, financial or safety contexts, the qualified domain owner and controlling documents determine the final deadline or permission, not a generic linguistic rule.

Questions readers ask

Is tense the same as event time?

No. Grammatical tense is an expression strategy relative to viewpoint; event time is the underlying situation. The reviewer should keep a record of the controlling evidence and the exact reader task used. This makes an apparently linguistic judgment reproducible and helps later editors avoid reintroducing the error.

Does past perfect need the same target verb form?

No. Preserve anteriority using natural target-language means. The reviewer should keep a record of the controlling evidence and the exact reader task used. This makes an apparently linguistic judgment reproducible and helps later editors avoid reintroducing the error.

Does has been working mean finished?

No. Duration of activity does not by itself certify completion. The reviewer should keep a record of the controlling evidence and the exact reader task used. This makes an apparently linguistic judgment reproducible and helps later editors avoid reintroducing the error.

Can by Friday and before Friday be substituted?

Not automatically. They can impose different boundaries. The reviewer should keep a record of the controlling evidence and the exact reader task used. This makes an apparently linguistic judgment reproducible and helps later editors avoid reintroducing the error.

Should a translator convert soon into an exact hour?

No. Keep the source’s precision unless an authorized clarification exists. The reviewer should keep a record of the controlling evidence and the exact reader task used. This makes an apparently linguistic judgment reproducible and helps later editors avoid reintroducing the error.

Does a UTC offset define a future local zone?

No. A named civil zone has time-dependent rules beyond an offset at one instant. The reviewer should keep a record of the controlling evidence and the exact reader task used. This makes an apparently linguistic judgment reproducible and helps later editors avoid reintroducing the error.

Can an archived tomorrow be read as tomorrow today?

No. It belongs to its original speaker or notice reference time. The reviewer should keep a record of the controlling evidence and the exact reader task used. This makes an apparently linguistic judgment reproducible and helps later editors avoid reintroducing the error.

How is a deadline tested?

Use the authorized instant, inclusion rule and sample actions immediately on both sides. The reviewer should keep a record of the controlling evidence and the exact reader task used. This makes an apparently linguistic judgment reproducible and helps later editors avoid reintroducing the error.

What if witnesses disagree about order?

Translate each attributed claim and preserve the uncertainty. The reviewer should keep a record of the controlling evidence and the exact reader task used. This makes an apparently linguistic judgment reproducible and helps later editors avoid reintroducing the error.

What if an AI answer is fluent but changes timing?

Treat it as semantic error and test event order and status explicitly. The reviewer should keep a record of the controlling evidence and the exact reader task used. This makes an apparently linguistic judgment reproducible and helps later editors avoid reintroducing the error.

What if the source lacks a year?

Do not invent it; request an authoritative anchor or preserve uncertainty. The reviewer should keep a record of the controlling evidence and the exact reader task used. This makes an apparently linguistic judgment reproducible and helps later editors avoid reintroducing the error.

What does a successful review demonstrate?

A reader of the target can reconstruct the source-supported sequence, state, precision and permissions. The reviewer should keep a record of the controlling evidence and the exact reader task used. This makes an apparently linguistic judgment reproducible and helps later editors avoid reintroducing the error.

High-authority references and their proper scope

ISO 24617-1:2012, SemAF-Time / ISO-TimeML: ISO’s official record establishes a semantic annotation approach to time and events and notes confirmation in 2023. It informs event-reference analysis but is not a table of target-language verb endings.

RFC 3339 and RFC 9557: RFC 3339, July 2002 and RFC 9557, April 2024 describe timestamp representation and further time-zone information. An offset is not a permanent named-zone rule. Read RFC 9557 together with the older RFC because it updates a particular offset interpretation.

Unicode CLDR / UTS #35 Part 4: Unicode dates specification and CLDR translation date-time pattern guidance cover the patterns, localized joins and relative-date forms used in software. Use these conventions for displaying known values, not for inventing author intent.

W3C Internationalization Tag Set 2.0: W3C ITS 2.0 provides translation/localization metadata concepts, including language information and localization notes. It supports carrying context through workflows; it does not independently prove the chronology of a particular human utterance.

Transfer practice: a repeatable set of reader tests

Choose one incident log, one story and one product notification. For each, draw two tables: chronological event order and order of presentation. Mark explicit completed, ongoing, repeated, proposed and uncertain states. Translate the passage in the normal target-language form and put the source away. Ask another proficient reader to rebuild the event graph and explain the action rule. Compare tables. Then change one source distinction, such as before to after or working to finished, and see whether the two outputs reliably differ. This develops a transferable skill rather than a memorized list of tense equivalences.

Use the method in production by linking the temporal contract to the source version and the approved translation. When a later correction arrives, test whether it changes event facts, knowledge state, effective dates or only publication wording. A new timestamp may be an edit record, not an event. A new fact may need an attributed explanatory note rather than retrospective alteration of an old quotation. Each release should let readers trace what was known, what was planned, what happened, and when the language describing it became authoritative.

Next routes within the translation architecture

This reserved-tens chapter follows Master Art of Translation | The Discourse Reference System, which tracks who or what a sentence refers to. Here the corresponding problem is when events occurred and whether they were finished, repeated, predicted or still unfolding. For the full global framework, return to Master Art of Translation | The Complete System for Moving Meaning Between Languages.

When the question is how to build lexical concepts rather than preserve cross-language event timing, use the protected Vocabulary Learning Hub. For additional supporting explanations of English and cross-language meaning, visit How English Works | Translation and Meaning Across Languages. Those resources remain under their existing owners. The continuing translation sequence can extend the same event-evidence discipline to other dimensions of meaning without changing the protected material.