How do you translate electric power grid operations, electricity outage notices, utility switching records and energy-management messages without changing what equipment is connected, what power is available, what caused an interruption, or what a customer should expect? Searches for power grid terminology translation, electricity distribution translation, SCADA alarm localization, outage restoration notices, switching-order translation and IEC power-system vocabulary all lead to one practical question: which facts must remain unchanged when the language changes? This guide treats translation as the controlled transfer of operational meaning, rather than the replacement of electrical words by apparently similar words in another language.
Technical translation in the electricity sector is difficult because the same short phrase can describe a physical condition, a sensor observation, an instruction awaiting approval, an intended future state or a public-facing summary. A breaker indicated as open is not necessarily proved isolated; an outage reported by a customer is not yet a diagnosed equipment fault; an estimated restoration time is not a guaranteed return of service; and a megawatt value is not interchangeable with a megawatt-hour reading. A fluent translation can be dangerous or commercially misleading if it silently changes any of those relationships. The challenge is to preserve asset identity, timing, units, uncertainty, authority, topology and evidence simultaneously.
This world-facing guide develops a repeatable method for translating electric utility records, control-centre narratives, protection events, network model labels, smart-meter communications, renewable-energy dispatch explanations, customer power outage updates and post-incident reports. It explains how to translate electricity network terms and grid restoration language accurately, how to distinguish IEC Common Information Model concepts from human-readable descriptions, how to review machine and AI translation of high-stakes technical text, and how to run bilingual quality checks against fictional worked scenarios. The reader will find mechanisms, a controlled production workflow, failure modes, diagnostics, repair exercises, a terminology glossary and routes to broader language learning. It does not provide operational permission to switch, isolate or energize electrical apparatus.
Architecture node: EDKSG-TRANS-MASTER-WORLD-560. Scope: the cross-language fidelity of electrical-operations information, not engineering design, regulatory approval, electrical safety training, or real-time switching instructions. Research reviewed: 30 October 2026. Every illustrative utility, asset code, reading, time, conversation and message is an invented teaching example unless expressly attributed to a published standard. Local law, utility operating rules and official emergency guidance always take precedence over an educational example.
The backbone of this branch is simple: establish what the source record asserts; identify whether each item is an object, observation, command, forecast, decision or explanatory phrase; classify which identifiers and measurements must be copied exactly; translate only the human language fields; compare the target against the source for scope and commitment; and obtain competent technical review before any safety-critical release. This prevents one article from claiming ownership of general electrical engineering, general vocabulary acquisition or English grammar. Those disciplines remain useful neighbours, while the present article owns their intersection with meaning moving between languages.
An electric grid is a set of connected technical and human systems, not a single machine with a single status. Generation, transmission, distribution, metering, demand, protection, field operations, customer communications and regulated records can describe different slices of the same event. A message saying a section is unavailable may be correct for a monitoring display but insufficient for a customer’s question; a crew note saying a fuse was replaced may be a completed task but not proof that supply has returned to every connected premise. Translation has to respect those differences instead of smoothing them into one attractive narrative.
Three families of reference material anchor terminology without replacing source analysis. The IEC International Electrotechnical Vocabulary, commonly available as Electropedia, supports concept-level electrical terminology. IEC 61970’s Common Information Model gives a standardised semantic representation for exchanging power-system resources and relationships; IEC 61968 addresses distribution-management interfaces, including network operations. IEC 61850 describes information models and communication structures used in utility automation. These materials are not universal bilingual word lists, and their existence does not make every privately written utility label a normative IEC term. The translator must verify the actual standard edition, system version and client-specific data dictionary.
Do not confuse operational accuracy with legal or safety approval. This guide will never instruct an unqualified reader to perform a switching sequence. When a source sentence contains a real-world directive, the translator preserves its status and sends uncertainty to the authorised engineer or operations owner. A translation does not create permission to operate equipment, does not clear a work permit, does not establish a safe isolation boundary, and does not constitute a dispatch order. The most responsible answer to an ambiguous command is a documented question, not a plausible guess.
To use the guide, first identify the document’s life cycle. A reference manual needs controlled terminology and reproducible revisions. A live alarm needs faithful states, timestamps and severity. A switching record needs exact actor, equipment, condition and authorisation. A customer notice needs comprehensible impacts, caveats and actions. A regulatory filing needs defensible source evidence, definitions and version history. The same bilingual term may be usable in one document class but inadequate in another. Beginning with the reader’s job prevents technical translation from turning into a dictionary competition.
Readers seeking basic learning of words should use the Vocabulary Learning Hub; readers needing English grammar or register inside English should use How English Works. Those owners remain separate. A learner who knows the words open, close, trip and restore may still need this specialist guide to determine when those words describe a circuit state, an automatic protective action, a work activity or the availability of service. The article is therefore an application and transfer layer, not a replacement for general language education.
Reader route: from search intent to the right translation decision
Start with the hidden problem if the target sentence sounds natural but might be wrong. Use the mechanism chapters when you are deciding which grid concepts can safely be paraphrased and which demand exact field-level fidelity. The repeatable method is for project managers and bilingual reviewers who need a process rather than ad hoc proofreading. The failure-mode section is for triage; the worked diagnostics let teams compare an invented source with a misleading target and a repaired rendition; the transfer section turns those repairs into deliberate practice.
Think in five linked layers. The language layer carries terminology, grammar, voice and audience. The model layer carries entities, identifiers and relations. The operational layer carries state, observation, instruction, permission and sequence. The evidential layer carries origin, timestamp, confidence and responsibility. The delivery layer carries interface limits, notification channels, accessibility and version control. Translation is adequate only when the target preserves the relevant constraints in all five layers.
Use the following question before every translation decision: would the intended reader take the same authorised action, believe the same bounded claim and identify the same underlying object after reading the target text? If the answer is uncertain, there is a problem to resolve. A literal rendering can fail this test; a polished free rendering can fail it too. The goal is neither word worship nor unlimited rewriting. It is traceable, purpose-sensitive equivalence.
An operational statement also has a time dimension. A status can be valid at the sampling instant, at the time a database was last updated, or at the moment a human published a notice. Those times can differ significantly. A translated message that drops its ‘as of’ clause may change a historical observation into an apparently live claim. This is a reasoning failure even when every electrical noun is translated correctly.
Numbers need their own types. Voltage, current, active power, reactive power, energy, duration, frequency, temperature, position and count each answer a different question. Their displayed numerals may look similar, but the suffixes and context determine interpretation. A good translator does not infer an omitted unit from familiarity, switch decimal conventions silently, or treat the customer-facing expression ‘power is back’ as a quantitative electrical measurement.
The hidden problem: a sentence is only one view of an operational state
Consider a fictional message: ‘Feeder F-17 open; restoration pending validation.’ Three kinds of information have been compressed into six words: a named network object, an observed or declared state, and a workflow condition that has not yet been satisfied. If a translation says the feeder has been safely isolated and electricity will return soon, it has added two assertions unsupported by the source. Safe isolation is an operational determination, while soon is a prediction. Neither is a synonym of open or pending validation.
Grid language contains many propositions of the form ‘X was true at time T according to source S’ rather than ‘X is always true’. Supervisory control and data acquisition, telemetry gateways, enterprise data platforms, field reports, smart meters, call-centre tickets and public service banners may be updated at different rates. Translators must identify the claim’s provenance before deciding whether to use a present tense, a completed aspect, or a future commitment in a target language.
Translation errors often hide in grammar rather than obvious vocabulary. A source may say that a circuit may be restored after checks, while a target says it will be restored after checks. A source may say a repair team has been assigned, while a target says repairs are complete. A source may use an indefinite count, while a target gives a definitive number. These are changes in modality, agency, completion and evidential strength; fluent wording cannot excuse them.
Another hidden problem is category collapse. A switch, breaker, fuse, recloser, protective relay, transformer, feeder and customer meter are not mutually replaceable nouns. Even where a general-audience notice uses ‘electrical equipment’, the controlled record must retain the correct asset and relation. Conversely, a public notice should not reproduce opaque asset identifiers without a usable explanation. The translator must know which audience receives which level of detail while preserving the same underlying truth.
Source ambiguity is not an invitation to invention. If a dataset uses ‘isolated’ inconsistently, the right response may be to pause that field, ask whether it means separated from a source of supply, administratively blocked from service, or another documented status, and record the approved interpretation. A reviewer should be able to follow the resolution back to the originating author. A hidden editorial correction that cannot be traced is not robust translation.
Emergency and customer communications deserve special care. A general statement that crews are assessing damage must not become advice to handle a downed conductor; a notice that the estimated restoration time is being revised must not turn into a guarantee. Translators should convey the source’s actionable safety warnings faithfully, preserve limitations and link to the issuing authority. They should not improvise electrical-safety procedures or supply personal instructions beyond the source.
A final hidden problem is the illusion of single-version truth. A translated outage banner may be cached while the source record is edited; a changed estimate may never reach a second-language channel; a previous advisory may remain prominent after it is superseded. Translation quality therefore has a release-engineering component: language variants must share stable record IDs, effective periods and retirement behaviour. Being correct yesterday is not the same as being current today.
Mechanism: thirty-two invariants that power-system translations must preserve
01. Grid entity identity
What must remain true. The asset code identifies an object, not a free-form description. In professional cross-language work, the central unit of analysis is not the isolated word but the claim carried by an identified source. A term may have a direct bilingual equivalent while still needing a time qualifier, data type or institutional role to become accurate. The translator should name the controlled object first, then decide how a target reader is expected to use the sentence.
Why fluency is not enough. A feeder name, equipment ID or station code may resemble a translatable noun; changing it would sever traceability. This is especially important where a source mixes engineering data with plain-language notes. A machine translation that expands a compact abbreviation into an attractive sentence can lose the distinction between observation, plan, possibility and confirmation. The correct response is to resolve the underlying semantics, not to hide uncertainty behind polished target-language style.
Illustrative diagnostic. In a fictional ticket ‘FDR-ALPHA-07’ is a controlled key; ‘north feeder’ is a human description and may be localized. The example is fictional and demonstrates an interpretation boundary rather than a live network condition. Freeze identifiers, translate the descriptive layer and show the relationship between both. Preserve the original record in a review ledger so that a second bilingual reviewer can evaluate the decision without relying on the translator’s memory.
Release test. The reviewer should be able to retrieve the identical record using either language without matching fuzzy text. Read the target without the source and ask what action, belief or inference it encourages. Read both versions together and check whether the same asset, time, amount, conditionality and responsibility survive. If the two tests disagree, do not approve the rendering until the specific gap is resolved. Keep the technical judgement with the authorised domain specialist and the linguistic judgement with the accountable translator.
02. Transmission and distribution boundaries
What must remain true. Network layers have different responsibilities and audiences. In professional cross-language work, the central unit of analysis is not the isolated word but the claim carried by an identified source. A term may have a direct bilingual equivalent while still needing a time qualifier, data type or institutional role to become accurate. The translator should name the controlled object first, then decide how a target reader is expected to use the sentence.
Why fluency is not enough. An operational sentence can refer to bulk transmission, a distribution feeder, service connection or behind-the-meter circuit. This is especially important where a source mixes engineering data with plain-language notes. A machine translation that expands a compact abbreviation into an attractive sentence can lose the distinction between observation, plan, possibility and confirmation. The correct response is to resolve the underlying semantics, not to hide uncertainty behind polished target-language style.
Illustrative diagnostic. ‘Transmission corridor constrained’ must not become ‘your home wiring failed’; the latter changes the responsible system. The example is fictional and demonstrates an interpretation boundary rather than a live network condition. Preserve level of network and the named operator; permit plain-language explanation without substituting an asset. Preserve the original record in a review ledger so that a second bilingual reviewer can evaluate the decision without relying on the translator’s memory.
Release test. Match the affected part of the network in a diagram or approved data dictionary. Read the target without the source and ask what action, belief or inference it encourages. Read both versions together and check whether the same asset, time, amount, conditionality and responsibility survive. If the two tests disagree, do not approve the rendering until the specific gap is resolved. Keep the technical judgement with the authorised domain specialist and the linguistic judgement with the accountable translator.
03. Feeder connectivity and topology
What must remain true. Connectivity describes how network objects relate at a particular model or observation time. In professional cross-language work, the central unit of analysis is not the isolated word but the claim carried by an identified source. A term may have a direct bilingual equivalent while still needing a time qualifier, data type or institutional role to become accurate. The translator should name the controlled object first, then decide how a target reader is expected to use the sentence.
Why fluency is not enough. The word connected may mean a physical conductor, an enabled network path, a calculated topology or an assumed model edge. This is especially important where a source mixes engineering data with plain-language notes. A machine translation that expands a compact abbreviation into an attractive sentence can lose the distinction between observation, plan, possibility and confirmation. The correct response is to resolve the underlying semantics, not to hide uncertainty behind polished target-language style.
Illustrative diagnostic. A feeder map marked as connected cannot by itself prove that a customer has energised supply. The example is fictional and demonstrates an interpretation boundary rather than a live network condition. Keep network-model status separate from live service status, and identify the source and time of the topology snapshot. Preserve the original record in a review ledger so that a second bilingual reviewer can evaluate the decision without relying on the translator’s memory.
Release test. Test with a counterexample in which the path exists but a protective device is open. Read the target without the source and ask what action, belief or inference it encourages. Read both versions together and check whether the same asset, time, amount, conditionality and responsibility survive. If the two tests disagree, do not approve the rendering until the specific gap is resolved. Keep the technical judgement with the authorised domain specialist and the linguistic judgement with the accountable translator.
04. Circuit-breaker position
What must remain true. Mechanical position, remote indication and authorised operating intent are separate concepts. In professional cross-language work, the central unit of analysis is not the isolated word but the claim carried by an identified source. A term may have a direct bilingual equivalent while still needing a time qualifier, data type or institutional role to become accurate. The translator should name the controlled object first, then decide how a target reader is expected to use the sentence.
Why fluency is not enough. Open and closed can mean contact position, telemetry value, commanded target or confirmed state. This is especially important where a source mixes engineering data with plain-language notes. A machine translation that expands a compact abbreviation into an attractive sentence can lose the distinction between observation, plan, possibility and confirmation. The correct response is to resolve the underlying semantics, not to hide uncertainty behind polished target-language style.
Illustrative diagnostic. ‘Breaker open indication received’ is not equivalent to ‘breaker opened safely on command’. The example is fictional and demonstrates an interpretation boundary rather than a live network condition. Tag verbs with observed, commanded or confirmed status before translating tense and agency. Preserve the original record in a review ledger so that a second bilingual reviewer can evaluate the decision without relying on the translator’s memory.
Release test. Compare source points and event logs rather than interpreting a single screen colour. Read the target without the source and ask what action, belief or inference it encourages. Read both versions together and check whether the same asset, time, amount, conditionality and responsibility survive. If the two tests disagree, do not approve the rendering until the specific gap is resolved. Keep the technical judgement with the authorised domain specialist and the linguistic judgement with the accountable translator.
05. Disconnectors and isolating devices
What must remain true. Device type is part of the technical claim. In professional cross-language work, the central unit of analysis is not the isolated word but the claim carried by an identified source. A term may have a direct bilingual equivalent while still needing a time qualifier, data type or institutional role to become accurate. The translator should name the controlled object first, then decide how a target reader is expected to use the sentence.
Why fluency is not enough. Generic words such as switch and isolator can conceal different functions, ratings and procedures. This is especially important where a source mixes engineering data with plain-language notes. A machine translation that expands a compact abbreviation into an attractive sentence can lose the distinction between observation, plan, possibility and confirmation. The correct response is to resolve the underlying semantics, not to hide uncertainty behind polished target-language style.
Illustrative diagnostic. A note describing an off-load disconnector cannot be rewritten as a circuit breaker trip. The example is fictional and demonstrates an interpretation boundary rather than a live network condition. Preserve named device class, and escalate a source using an unapproved generic term. Preserve the original record in a review ledger so that a second bilingual reviewer can evaluate the decision without relying on the translator’s memory.
Release test. Check equipment taxonomy and the utility’s controlled symbol legend. Read the target without the source and ask what action, belief or inference it encourages. Read both versions together and check whether the same asset, time, amount, conditionality and responsibility survive. If the two tests disagree, do not approve the rendering until the specific gap is resolved. Keep the technical judgement with the authorised domain specialist and the linguistic judgement with the accountable translator.
06. Energized, de-energized and proved dead
What must remain true. These phrases do not all establish the same electrical-safety condition. In professional cross-language work, the central unit of analysis is not the isolated word but the claim carried by an identified source. A term may have a direct bilingual equivalent while still needing a time qualifier, data type or institutional role to become accurate. The translator should name the controlled object first, then decide how a target reader is expected to use the sentence.
Why fluency is not enough. A system can be reported not energised without a completed safe-work verification. This is especially important where a source mixes engineering data with plain-language notes. A machine translation that expands a compact abbreviation into an attractive sentence can lose the distinction between observation, plan, possibility and confirmation. The correct response is to resolve the underlying semantics, not to hide uncertainty behind polished target-language style.
Illustrative diagnostic. ‘No voltage detected’ is a measurement statement, not an automatic permit to work. The example is fictional and demonstrates an interpretation boundary rather than a live network condition. Translate only the source assertion, its measurement scope and evidentiary status. Preserve the original record in a review ledger so that a second bilingual reviewer can evaluate the decision without relying on the translator’s memory.
Release test. Require the authorised safety owner to approve any work-status rendering; this guide gives no switching instruction. Read the target without the source and ask what action, belief or inference it encourages. Read both versions together and check whether the same asset, time, amount, conditionality and responsibility survive. If the two tests disagree, do not approve the rendering until the specific gap is resolved. Keep the technical judgement with the authorised domain specialist and the linguistic judgement with the accountable translator.
07. Voltage magnitude and nominal class
What must remain true. Values carry physical units, reference conditions and uncertainty. In professional cross-language work, the central unit of analysis is not the isolated word but the claim carried by an identified source. A term may have a direct bilingual equivalent while still needing a time qualifier, data type or institutional role to become accurate. The translator should name the controlled object first, then decide how a target reader is expected to use the sentence.
Why fluency is not enough. A 11 kV network class, a measured 10.7 kV value and a per-unit voltage are not interchangeable. This is especially important where a source mixes engineering data with plain-language notes. A machine translation that expands a compact abbreviation into an attractive sentence can lose the distinction between observation, plan, possibility and confirmation. The correct response is to resolve the underlying semantics, not to hide uncertainty behind polished target-language style.
Illustrative diagnostic. A bilingual summary reporting ‘11,000 V nominal’ must not say ‘exactly 11,000 V measured’. The example is fictional and demonstrates an interpretation boundary rather than a live network condition. Preserve class versus measurement and retain engineering units and significant figures. Preserve the original record in a review ledger so that a second bilingual reviewer can evaluate the decision without relying on the translator’s memory.
Release test. Compare the numerical value, unit, phase reference and measurement timestamp. Read the target without the source and ask what action, belief or inference it encourages. Read both versions together and check whether the same asset, time, amount, conditionality and responsibility survive. If the two tests disagree, do not approve the rendering until the specific gap is resolved. Keep the technical judgement with the authorised domain specialist and the linguistic judgement with the accountable translator.
08. Active power versus electrical energy
What must remain true. Power is a rate; energy accumulates over time. In professional cross-language work, the central unit of analysis is not the isolated word but the claim carried by an identified source. A term may have a direct bilingual equivalent while still needing a time qualifier, data type or institutional role to become accurate. The translator should name the controlled object first, then decide how a target reader is expected to use the sentence.
Why fluency is not enough. MW and MWh appear one letter apart and can be mistranslated as the same ‘electricity’ quantity. This is especially important where a source mixes engineering data with plain-language notes. A machine translation that expands a compact abbreviation into an attractive sentence can lose the distinction between observation, plan, possibility and confirmation. The correct response is to resolve the underlying semantics, not to hide uncertainty behind polished target-language style.
Illustrative diagnostic. ‘Peak demand 42 MW’ does not report consumption of 42 MWh. The example is fictional and demonstrates an interpretation boundary rather than a live network condition. Use concept-level terminology and check the quantity definition before any locale conversion. Preserve the original record in a review ledger so that a second bilingual reviewer can evaluate the decision without relying on the translator’s memory.
Release test. Verify that the time interval is present for consumption and that values remain typed. Read the target without the source and ask what action, belief or inference it encourages. Read both versions together and check whether the same asset, time, amount, conditionality and responsibility survive. If the two tests disagree, do not approve the rendering until the specific gap is resolved. Keep the technical judgement with the authorised domain specialist and the linguistic judgement with the accountable translator.
09. Reactive power and power factor
What must remain true. Electrical-system performance measures have distinct definitions. In professional cross-language work, the central unit of analysis is not the isolated word but the claim carried by an identified source. A term may have a direct bilingual equivalent while still needing a time qualifier, data type or institutional role to become accurate. The translator should name the controlled object first, then decide how a target reader is expected to use the sentence.
Why fluency is not enough. Mvar, MVA, MW and a dimensionless power factor are easily flattened into ‘power’. This is especially important where a source mixes engineering data with plain-language notes. A machine translation that expands a compact abbreviation into an attractive sentence can lose the distinction between observation, plan, possibility and confirmation. The correct response is to resolve the underlying semantics, not to hide uncertainty behind polished target-language style.
Illustrative diagnostic. A capacitor bank providing reactive support is not generating 6 MW of real energy. The example is fictional and demonstrates an interpretation boundary rather than a live network condition. Leave units intact, explain the concept only when the audience requires it, and avoid adding causal claims. Preserve the original record in a review ledger so that a second bilingual reviewer can evaluate the decision without relying on the translator’s memory.
Release test. Cross-check quantity symbols and whether a measurement is leading, lagging or sign-convention dependent. Read the target without the source and ask what action, belief or inference it encourages. Read both versions together and check whether the same asset, time, amount, conditionality and responsibility survive. If the two tests disagree, do not approve the rendering until the specific gap is resolved. Keep the technical judgement with the authorised domain specialist and the linguistic judgement with the accountable translator.
10. Frequency and control response
What must remain true. Frequency is an observable, while control actions and reserves are mechanisms. In professional cross-language work, the central unit of analysis is not the isolated word but the claim carried by an identified source. A term may have a direct bilingual equivalent while still needing a time qualifier, data type or institutional role to become accurate. The translator should name the controlled object first, then decide how a target reader is expected to use the sentence.
Why fluency is not enough. A 49.9 Hz observation should not become an announcement that a generator was dispatched. This is especially important where a source mixes engineering data with plain-language notes. A machine translation that expands a compact abbreviation into an attractive sentence can lose the distinction between observation, plan, possibility and confirmation. The correct response is to resolve the underlying semantics, not to hide uncertainty behind polished target-language style.
Illustrative diagnostic. ‘Frequency recovered’ may describe the measured value, not proof that every customer has supply. The example is fictional and demonstrates an interpretation boundary rather than a live network condition. Keep measured outcome distinct from control action and geographical scope. Preserve the original record in a review ledger so that a second bilingual reviewer can evaluate the decision without relying on the translator’s memory.
Release test. Record observation period, measurement point and the condition used to judge recovery. Read the target without the source and ask what action, belief or inference it encourages. Read both versions together and check whether the same asset, time, amount, conditionality and responsibility survive. If the two tests disagree, do not approve the rendering until the specific gap is resolved. Keep the technical judgement with the authorised domain specialist and the linguistic judgement with the accountable translator.
11. Generation dispatch and availability
What must remain true. Available capacity, scheduled output, actual generation and instructed setpoints differ. In professional cross-language work, the central unit of analysis is not the isolated word but the claim carried by an identified source. A term may have a direct bilingual equivalent while still needing a time qualifier, data type or institutional role to become accurate. The translator should name the controlled object first, then decide how a target reader is expected to use the sentence.
Why fluency is not enough. ‘Available’ is often read casually as ‘currently producing’. This is especially important where a source mixes engineering data with plain-language notes. A machine translation that expands a compact abbreviation into an attractive sentence can lose the distinction between observation, plan, possibility and confirmation. The correct response is to resolve the underlying semantics, not to hide uncertainty behind polished target-language style.
Illustrative diagnostic. A plant with 100 MW available capacity but 60 MW scheduled is not necessarily outputting 100 MW. The example is fictional and demonstrates an interpretation boundary rather than a live network condition. Translate status taxonomy first; preserve setpoint, actual value and applicable time separately. Preserve the original record in a review ledger so that a second bilingual reviewer can evaluate the decision without relying on the translator’s memory.
Release test. Require the reviewer to reconstruct a timeline without guessing from a single number. Read the target without the source and ask what action, belief or inference it encourages. Read both versions together and check whether the same asset, time, amount, conditionality and responsibility survive. If the two tests disagree, do not approve the rendering until the specific gap is resolved. Keep the technical judgement with the authorised domain specialist and the linguistic judgement with the accountable translator.
12. Load-shedding and curtailment
What must remain true. Reducing demand is not the same as constraining generation. In professional cross-language work, the central unit of analysis is not the isolated word but the claim carried by an identified source. A term may have a direct bilingual equivalent while still needing a time qualifier, data type or institutional role to become accurate. The translator should name the controlled object first, then decide how a target reader is expected to use the sentence.
Why fluency is not enough. The verbs cut, shed, curtail and disconnect have different objects and causes. This is especially important where a source mixes engineering data with plain-language notes. A machine translation that expands a compact abbreviation into an attractive sentence can lose the distinction between observation, plan, possibility and confirmation. The correct response is to resolve the underlying semantics, not to hide uncertainty behind polished target-language style.
Illustrative diagnostic. ‘Solar output curtailed’ must not become ‘customers shed load’. The example is fictional and demonstrates an interpretation boundary rather than a live network condition. Identify the actor, affected equipment, intended variable, duration and authorisation. Preserve the original record in a review ledger so that a second bilingual reviewer can evaluate the decision without relying on the translator’s memory.
Release test. Check whether the source describes forced, scheduled, requested or voluntary reduction. Read the target without the source and ask what action, belief or inference it encourages. Read both versions together and check whether the same asset, time, amount, conditionality and responsibility survive. If the two tests disagree, do not approve the rendering until the specific gap is resolved. Keep the technical judgement with the authorised domain specialist and the linguistic judgement with the accountable translator.
13. Transformer taps and voltage regulation
What must remain true. Tap position is a device setting, not an unrestricted quality adjective. In professional cross-language work, the central unit of analysis is not the isolated word but the claim carried by an identified source. A term may have a direct bilingual equivalent while still needing a time qualifier, data type or institutional role to become accurate. The translator should name the controlled object first, then decide how a target reader is expected to use the sentence.
Why fluency is not enough. A translation may omit whether an adjustment was commanded, observed or blocked. This is especially important where a source mixes engineering data with plain-language notes. A machine translation that expands a compact abbreviation into an attractive sentence can lose the distinction between observation, plan, possibility and confirmation. The correct response is to resolve the underlying semantics, not to hide uncertainty behind polished target-language style.
Illustrative diagnostic. ‘Tap request queued’ must not become ‘voltage regulation completed’. The example is fictional and demonstrates an interpretation boundary rather than a live network condition. Keep target position, actual position and change confirmation separate. Preserve the original record in a review ledger so that a second bilingual reviewer can evaluate the decision without relying on the translator’s memory.
Release test. Inspect source fields for request status and telemetered feedback. Read the target without the source and ask what action, belief or inference it encourages. Read both versions together and check whether the same asset, time, amount, conditionality and responsibility survive. If the two tests disagree, do not approve the rendering until the specific gap is resolved. Keep the technical judgement with the authorised domain specialist and the linguistic judgement with the accountable translator.
14. Protection trip versus manual opening
What must remain true. An automatic protection event and a human-initiated operation are different causes. In professional cross-language work, the central unit of analysis is not the isolated word but the claim carried by an identified source. A term may have a direct bilingual equivalent while still needing a time qualifier, data type or institutional role to become accurate. The translator should name the controlled object first, then decide how a target reader is expected to use the sentence.
Why fluency is not enough. The everyday verb opened hides whether a breaker tripped or was deliberately opened. This is especially important where a source mixes engineering data with plain-language notes. A machine translation that expands a compact abbreviation into an attractive sentence can lose the distinction between observation, plan, possibility and confirmation. The correct response is to resolve the underlying semantics, not to hide uncertainty behind polished target-language style.
Illustrative diagnostic. ‘Trip asserted’ does not prove a fault type or an operator’s intention. The example is fictional and demonstrates an interpretation boundary rather than a live network condition. Preserve event category, protective function and attribution boundary. Preserve the original record in a review ledger so that a second bilingual reviewer can evaluate the decision without relying on the translator’s memory.
Release test. Correlate available event codes without inventing a root-cause diagnosis. Read the target without the source and ask what action, belief or inference it encourages. Read both versions together and check whether the same asset, time, amount, conditionality and responsibility survive. If the two tests disagree, do not approve the rendering until the specific gap is resolved. Keep the technical judgement with the authorised domain specialist and the linguistic judgement with the accountable translator.
15. Fault detection versus root cause
What must remain true. Indications, alarms and preliminary classifications are not investigations. In professional cross-language work, the central unit of analysis is not the isolated word but the claim carried by an identified source. A term may have a direct bilingual equivalent while still needing a time qualifier, data type or institutional role to become accurate. The translator should name the controlled object first, then decide how a target reader is expected to use the sentence.
Why fluency is not enough. A relay flag may indicate an abnormal condition without identifying a damaged component. This is especially important where a source mixes engineering data with plain-language notes. A machine translation that expands a compact abbreviation into an attractive sentence can lose the distinction between observation, plan, possibility and confirmation. The correct response is to resolve the underlying semantics, not to hide uncertainty behind polished target-language style.
Illustrative diagnostic. ‘Earth-fault element operated’ must not be converted into ‘underground cable damaged’. The example is fictional and demonstrates an interpretation boundary rather than a live network condition. Retain observed evidence and state explicit uncertainty about causal conclusions. Preserve the original record in a review ledger so that a second bilingual reviewer can evaluate the decision without relying on the translator’s memory.
Release test. Check whether a signed investigation or field inspection establishes the cause. Read the target without the source and ask what action, belief or inference it encourages. Read both versions together and check whether the same asset, time, amount, conditionality and responsibility survive. If the two tests disagree, do not approve the rendering until the specific gap is resolved. Keep the technical judgement with the authorised domain specialist and the linguistic judgement with the accountable translator.
16. Fault location uncertainty
What must remain true. Estimated location, confirmed defect site and affected service area are separate. In professional cross-language work, the central unit of analysis is not the isolated word but the claim carried by an identified source. A term may have a direct bilingual equivalent while still needing a time qualifier, data type or institutional role to become accurate. The translator should name the controlled object first, then decide how a target reader is expected to use the sentence.
Why fluency is not enough. A model’s estimated section can look like a precise location in a public message. This is especially important where a source mixes engineering data with plain-language notes. A machine translation that expands a compact abbreviation into an attractive sentence can lose the distinction between observation, plan, possibility and confirmation. The correct response is to resolve the underlying semantics, not to hide uncertainty behind polished target-language style.
Illustrative diagnostic. ‘Suspected segment between P3 and P4’ does not establish the fault at pole P4. The example is fictional and demonstrates an interpretation boundary rather than a live network condition. Retain the qualifier, spatial scope and confidence source; never invent street-level detail. Preserve the original record in a review ledger so that a second bilingual reviewer can evaluate the decision without relying on the translator’s memory.
Release test. Review map geometry against the approved asset records and field confirmation. Read the target without the source and ask what action, belief or inference it encourages. Read both versions together and check whether the same asset, time, amount, conditionality and responsibility survive. If the two tests disagree, do not approve the rendering until the specific gap is resolved. Keep the technical judgement with the authorised domain specialist and the linguistic judgement with the accountable translator.
17. Planned work versus unplanned outage
What must remain true. The planning classification changes customer expectations and obligations. In professional cross-language work, the central unit of analysis is not the isolated word but the claim carried by an identified source. A term may have a direct bilingual equivalent while still needing a time qualifier, data type or institutional role to become accurate. The translator should name the controlled object first, then decide how a target reader is expected to use the sentence.
Why fluency is not enough. An interruption can follow scheduled maintenance, equipment failure, weather or safety response. This is especially important where a source mixes engineering data with plain-language notes. A machine translation that expands a compact abbreviation into an attractive sentence can lose the distinction between observation, plan, possibility and confirmation. The correct response is to resolve the underlying semantics, not to hide uncertainty behind polished target-language style.
Illustrative diagnostic. ‘Planned isolation pending approval’ is not an announced customer power cut. The example is fictional and demonstrates an interpretation boundary rather than a live network condition. Translate plan status, effective time and affected audience distinctly. Preserve the original record in a review ledger so that a second bilingual reviewer can evaluate the decision without relying on the translator’s memory.
Release test. Verify whether work has been authorised, notified, started, cancelled or completed. Read the target without the source and ask what action, belief or inference it encourages. Read both versions together and check whether the same asset, time, amount, conditionality and responsibility survive. If the two tests disagree, do not approve the rendering until the specific gap is resolved. Keep the technical judgement with the authorised domain specialist and the linguistic judgement with the accountable translator.
18. Restoration estimate versus commitment
What must remain true. An estimated time of restoration is a forecast with revision risk. In professional cross-language work, the central unit of analysis is not the isolated word but the claim carried by an identified source. A term may have a direct bilingual equivalent while still needing a time qualifier, data type or institutional role to become accurate. The translator should name the controlled object first, then decide how a target reader is expected to use the sentence.
Why fluency is not enough. Target-language future tense may sound guaranteed when the source is only an estimate. This is especially important where a source mixes engineering data with plain-language notes. A machine translation that expands a compact abbreviation into an attractive sentence can lose the distinction between observation, plan, possibility and confirmation. The correct response is to resolve the underlying semantics, not to hide uncertainty behind polished target-language style.
Illustrative diagnostic. ‘Estimated restoration after 18:00’ must not say ‘power will return at 18:00’. The example is fictional and demonstrates an interpretation boundary rather than a live network condition. Keep ‘estimated’, update time, dependency and uncertainty visible. Preserve the original record in a review ledger so that a second bilingual reviewer can evaluate the decision without relying on the translator’s memory.
Release test. Check public-facing copy for promised outcomes unsupported by the source. Read the target without the source and ask what action, belief or inference it encourages. Read both versions together and check whether the same asset, time, amount, conditionality and responsibility survive. If the two tests disagree, do not approve the rendering until the specific gap is resolved. Keep the technical judgement with the authorised domain specialist and the linguistic judgement with the accountable translator.
19. Partial restoration
What must remain true. Some customers can regain supply while others remain interrupted. In professional cross-language work, the central unit of analysis is not the isolated word but the claim carried by an identified source. A term may have a direct bilingual equivalent while still needing a time qualifier, data type or institutional role to become accurate. The translator should name the controlled object first, then decide how a target reader is expected to use the sentence.
Why fluency is not enough. A system-level flag saying restored may mask residual affected locations. This is especially important where a source mixes engineering data with plain-language notes. A machine translation that expands a compact abbreviation into an attractive sentence can lose the distinction between observation, plan, possibility and confirmation. The correct response is to resolve the underlying semantics, not to hide uncertainty behind polished target-language style.
Illustrative diagnostic. ‘Three of four zones restored’ cannot be reduced to ‘service fully restored’. The example is fictional and demonstrates an interpretation boundary rather than a live network condition. Preserve denominators, coverage and exceptions in every language. Preserve the original record in a review ledger so that a second bilingual reviewer can evaluate the decision without relying on the translator’s memory.
Release test. Compare restored counts against affected counts and latest incident scope. Read the target without the source and ask what action, belief or inference it encourages. Read both versions together and check whether the same asset, time, amount, conditionality and responsibility survive. If the two tests disagree, do not approve the rendering until the specific gap is resolved. Keep the technical judgement with the authorised domain specialist and the linguistic judgement with the accountable translator.
20. Field crews and work permits
What must remain true. Dispatch, arrival, assessment, repair and clearance are separate stages. In professional cross-language work, the central unit of analysis is not the isolated word but the claim carried by an identified source. A term may have a direct bilingual equivalent while still needing a time qualifier, data type or institutional role to become accurate. The translator should name the controlled object first, then decide how a target reader is expected to use the sentence.
Why fluency is not enough. An assigned crew is not necessarily onsite, and completed repairs do not equal a safety clearance. This is especially important where a source mixes engineering data with plain-language notes. A machine translation that expands a compact abbreviation into an attractive sentence can lose the distinction between observation, plan, possibility and confirmation. The correct response is to resolve the underlying semantics, not to hide uncertainty behind polished target-language style.
Illustrative diagnostic. ‘Crew en route’ must not become ‘crew repairing the fault now’. The example is fictional and demonstrates an interpretation boundary rather than a live network condition. Use exact workflow verbs, agent and checkpoint; do not turn translation into an operational authorization. Preserve the original record in a review ledger so that a second bilingual reviewer can evaluate the decision without relying on the translator’s memory.
Release test. Inspect timestamped crew status and approval records. Read the target without the source and ask what action, belief or inference it encourages. Read both versions together and check whether the same asset, time, amount, conditionality and responsibility survive. If the two tests disagree, do not approve the rendering until the specific gap is resolved. Keep the technical judgement with the authorised domain specialist and the linguistic judgement with the accountable translator.
21. SCADA measurement, indication and command
What must remain true. One screen can mix read-only telemetry and control requests. In professional cross-language work, the central unit of analysis is not the isolated word but the claim carried by an identified source. A term may have a direct bilingual equivalent while still needing a time qualifier, data type or institutional role to become accurate. The translator should name the controlled object first, then decide how a target reader is expected to use the sentence.
Why fluency is not enough. A label may look like a fact even when it represents a pending command or stale indication. This is especially important where a source mixes engineering data with plain-language notes. A machine translation that expands a compact abbreviation into an attractive sentence can lose the distinction between observation, plan, possibility and confirmation. The correct response is to resolve the underlying semantics, not to hide uncertainty behind polished target-language style.
Illustrative diagnostic. ‘Close request sent’ must not become ‘breaker closed’. The example is fictional and demonstrates an interpretation boundary rather than a live network condition. Mark record type and communication outcome before translating. Preserve the original record in a review ledger so that a second bilingual reviewer can evaluate the decision without relying on the translator’s memory.
Release test. Use the event chain to distinguish request, acknowledgement, execution and feedback. Read the target without the source and ask what action, belief or inference it encourages. Read both versions together and check whether the same asset, time, amount, conditionality and responsibility survive. If the two tests disagree, do not approve the rendering until the specific gap is resolved. Keep the technical judgement with the authorised domain specialist and the linguistic judgement with the accountable translator.
22. Stale, missing and invalid data
What must remain true. A value’s quality flag can be more important than the visible number. In professional cross-language work, the central unit of analysis is not the isolated word but the claim carried by an identified source. A term may have a direct bilingual equivalent while still needing a time qualifier, data type or institutional role to become accurate. The translator should name the controlled object first, then decide how a target reader is expected to use the sentence.
Why fluency is not enough. Last-known-good data may remain displayed after communication fails. This is especially important where a source mixes engineering data with plain-language notes. A machine translation that expands a compact abbreviation into an attractive sentence can lose the distinction between observation, plan, possibility and confirmation. The correct response is to resolve the underlying semantics, not to hide uncertainty behind polished target-language style.
Illustrative diagnostic. ‘Voltage 11.1 kV, quality stale’ is not a live 11.1 kV reading. The example is fictional and demonstrates an interpretation boundary rather than a live network condition. Protect quality markers and do not silently fill gaps with previous values. Preserve the original record in a review ledger so that a second bilingual reviewer can evaluate the decision without relying on the translator’s memory.
Release test. Inspect sample age, point validity and source-system quality fields. Read the target without the source and ask what action, belief or inference it encourages. Read both versions together and check whether the same asset, time, amount, conditionality and responsibility survive. If the two tests disagree, do not approve the rendering until the specific gap is resolved. Keep the technical judgement with the authorised domain specialist and the linguistic judgement with the accountable translator.
23. CIM identity and relationships
What must remain true. IEC 61970 CIM models entities and their connections for exchange. In professional cross-language work, the central unit of analysis is not the isolated word but the claim carried by an identified source. A term may have a direct bilingual equivalent while still needing a time qualifier, data type or institutional role to become accurate. The translator should name the controlled object first, then decide how a target reader is expected to use the sentence.
Why fluency is not enough. A human translation must not rename schema classes, keys or controlled attributes. This is especially important where a source mixes engineering data with plain-language notes. A machine translation that expands a compact abbreviation into an attractive sentence can lose the distinction between observation, plan, possibility and confirmation. The correct response is to resolve the underlying semantics, not to hide uncertainty behind polished target-language style.
Illustrative diagnostic. A user-facing ‘electrical bay’ label can change, but an API field key should remain stable. The example is fictional and demonstrates an interpretation boundary rather than a live network condition. Separate user-visible descriptions from interoperable data elements. Preserve the original record in a review ledger so that a second bilingual reviewer can evaluate the decision without relying on the translator’s memory.
Release test. Round-trip a sample dataset and compare object identity and relations. Read the target without the source and ask what action, belief or inference it encourages. Read both versions together and check whether the same asset, time, amount, conditionality and responsibility survive. If the two tests disagree, do not approve the rendering until the specific gap is resolved. Keep the technical judgement with the authorised domain specialist and the linguistic judgement with the accountable translator.
24. IEC 61850 logical nodes
What must remain true. Automation data models use standard logical-node and data-object names. In professional cross-language work, the central unit of analysis is not the isolated word but the claim carried by an identified source. A term may have a direct bilingual equivalent while still needing a time qualifier, data type or institutional role to become accurate. The translator should name the controlled object first, then decide how a target reader is expected to use the sentence.
Why fluency is not enough. Names that resemble English abbreviations often function as machine identifiers. This is especially important where a source mixes engineering data with plain-language notes. A machine translation that expands a compact abbreviation into an attractive sentence can lose the distinction between observation, plan, possibility and confirmation. The correct response is to resolve the underlying semantics, not to hide uncertainty behind polished target-language style.
Illustrative diagnostic. A logical-node identifier in a device message is not prose to be freely rewritten. The example is fictional and demonstrates an interpretation boundary rather than a live network condition. Keep code identifiers and machine-readable fields byte-stable unless formally mapped by engineering owners. Preserve the original record in a review ledger so that a second bilingual reviewer can evaluate the decision without relying on the translator’s memory.
Release test. Validate against the installed profile and declared standard edition. Read the target without the source and ask what action, belief or inference it encourages. Read both versions together and check whether the same asset, time, amount, conditionality and responsibility survive. If the two tests disagree, do not approve the rendering until the specific gap is resolved. Keep the technical judgement with the authorised domain specialist and the linguistic judgement with the accountable translator.
25. Meter intervals and consumption
What must remain true. Cumulative registers and interval readings represent different statistics. In professional cross-language work, the central unit of analysis is not the isolated word but the claim carried by an identified source. A term may have a direct bilingual equivalent while still needing a time qualifier, data type or institutional role to become accurate. The translator should name the controlled object first, then decide how a target reader is expected to use the sentence.
Why fluency is not enough. A reading may be in kWh, an interval usage quantity, or a device state. This is especially important where a source mixes engineering data with plain-language notes. A machine translation that expands a compact abbreviation into an attractive sentence can lose the distinction between observation, plan, possibility and confirmation. The correct response is to resolve the underlying semantics, not to hide uncertainty behind polished target-language style.
Illustrative diagnostic. ‘Interval energy 2.3 kWh’ is not equivalent to ‘current load 2.3 kW’. The example is fictional and demonstrates an interpretation boundary rather than a live network condition. Protect interval start, end, unit, timezone and measurement type. Preserve the original record in a review ledger so that a second bilingual reviewer can evaluate the decision without relying on the translator’s memory.
Release test. Reconcile against the original meter dataset and billing boundary. Read the target without the source and ask what action, belief or inference it encourages. Read both versions together and check whether the same asset, time, amount, conditionality and responsibility survive. If the two tests disagree, do not approve the rendering until the specific gap is resolved. Keep the technical judgement with the authorised domain specialist and the linguistic judgement with the accountable translator.
26. Distributed energy resources
What must remain true. Generation, storage and controllable demand may cross customer and grid boundaries. In professional cross-language work, the central unit of analysis is not the isolated word but the claim carried by an identified source. A term may have a direct bilingual equivalent while still needing a time qualifier, data type or institutional role to become accurate. The translator should name the controlled object first, then decide how a target reader is expected to use the sentence.
Why fluency is not enough. Export limits, installed capacity, actual power and stored energy may be confused. This is especially important where a source mixes engineering data with plain-language notes. A machine translation that expands a compact abbreviation into an attractive sentence can lose the distinction between observation, plan, possibility and confirmation. The correct response is to resolve the underlying semantics, not to hide uncertainty behind polished target-language style.
Illustrative diagnostic. A battery rated 8 MWh does not promise 8 MW instantaneous output. The example is fictional and demonstrates an interpretation boundary rather than a live network condition. Name the variable and the role of the device before choosing target words. Preserve the original record in a review ledger so that a second bilingual reviewer can evaluate the decision without relying on the translator’s memory.
Release test. Check whether the statement concerns interconnection permission, telemetry or commercial availability. Read the target without the source and ask what action, belief or inference it encourages. Read both versions together and check whether the same asset, time, amount, conditionality and responsibility survive. If the two tests disagree, do not approve the rendering until the specific gap is resolved. Keep the technical judgement with the authorised domain specialist and the linguistic judgement with the accountable translator.
27. Renewable forecasts and curtailment
What must remain true. Forecast production differs from actual output and accepted instructions. In professional cross-language work, the central unit of analysis is not the isolated word but the claim carried by an identified source. A term may have a direct bilingual equivalent while still needing a time qualifier, data type or institutional role to become accurate. The translator should name the controlled object first, then decide how a target reader is expected to use the sentence.
Why fluency is not enough. Probability bands can be removed when translating a concise dashboard. This is especially important where a source mixes engineering data with plain-language notes. A machine translation that expands a compact abbreviation into an attractive sentence can lose the distinction between observation, plan, possibility and confirmation. The correct response is to resolve the underlying semantics, not to hide uncertainty behind polished target-language style.
Illustrative diagnostic. ‘Forecast 75 MW ±10 MW’ is not a guaranteed output of 75 MW. The example is fictional and demonstrates an interpretation boundary rather than a live network condition. Preserve the prediction horizon, methodology status and uncertainty. Preserve the original record in a review ledger so that a second bilingual reviewer can evaluate the decision without relying on the translator’s memory.
Release test. Compare scenario bands, forecast issue times and reported actuals. Read the target without the source and ask what action, belief or inference it encourages. Read both versions together and check whether the same asset, time, amount, conditionality and responsibility survive. If the two tests disagree, do not approve the rendering until the specific gap is resolved. Keep the technical judgement with the authorised domain specialist and the linguistic judgement with the accountable translator.
28. Black-start and system restoration phases
What must remain true. System restoration uses technical states and approved responsibilities. In professional cross-language work, the central unit of analysis is not the isolated word but the claim carried by an identified source. A term may have a direct bilingual equivalent while still needing a time qualifier, data type or institutional role to become accurate. The translator should name the controlled object first, then decide how a target reader is expected to use the sentence.
Why fluency is not enough. ‘Black-start capable’ is a capability claim, not proof a plant has started. This is especially important where a source mixes engineering data with plain-language notes. A machine translation that expands a compact abbreviation into an attractive sentence can lose the distinction between observation, plan, possibility and confirmation. The correct response is to resolve the underlying semantics, not to hide uncertainty behind polished target-language style.
Illustrative diagnostic. ‘Restoration sequence under review’ is not an instruction that a grid is ready to energize. The example is fictional and demonstrates an interpretation boundary rather than a live network condition. Retain phase names as validated terms and avoid publishing actionable procedure steps. Preserve the original record in a review ledger so that a second bilingual reviewer can evaluate the decision without relying on the translator’s memory.
Release test. Obtain authorised technical review for operationally sensitive source passages. Read the target without the source and ask what action, belief or inference it encourages. Read both versions together and check whether the same asset, time, amount, conditionality and responsibility survive. If the two tests disagree, do not approve the rendering until the specific gap is resolved. Keep the technical judgement with the authorised domain specialist and the linguistic judgement with the accountable translator.
29. Demand-response and customer participation
What must remain true. A request to reduce demand is not the same as enforced interruption. In professional cross-language work, the central unit of analysis is not the isolated word but the claim carried by an identified source. A term may have a direct bilingual equivalent while still needing a time qualifier, data type or institutional role to become accurate. The translator should name the controlled object first, then decide how a target reader is expected to use the sentence.
Why fluency is not enough. Opt-in, programme eligibility, target reduction and actual response are separate. This is especially important where a source mixes engineering data with plain-language notes. A machine translation that expands a compact abbreviation into an attractive sentence can lose the distinction between observation, plan, possibility and confirmation. The correct response is to resolve the underlying semantics, not to hide uncertainty behind polished target-language style.
Illustrative diagnostic. ‘Eligible participant invited’ cannot become ‘customer enrolled and required’. The example is fictional and demonstrates an interpretation boundary rather than a live network condition. Preserve consent, obligation, reward and notice timing. Preserve the original record in a review ledger so that a second bilingual reviewer can evaluate the decision without relying on the translator’s memory.
Release test. Audit what a reader might think they must do and whether the source authorised that interpretation. Read the target without the source and ask what action, belief or inference it encourages. Read both versions together and check whether the same asset, time, amount, conditionality and responsibility survive. If the two tests disagree, do not approve the rendering until the specific gap is resolved. Keep the technical judgement with the authorised domain specialist and the linguistic judgement with the accountable translator.
30. Outage tickets and customer-facing updates
What must remain true. Internal troubleshooting and public advice are different information products. In professional cross-language work, the central unit of analysis is not the isolated word but the claim carried by an identified source. A term may have a direct bilingual equivalent while still needing a time qualifier, data type or institutional role to become accurate. The translator should name the controlled object first, then decide how a target reader is expected to use the sentence.
Why fluency is not enough. An internal cause hypothesis can become an unjustified public fact through simplification. This is especially important where a source mixes engineering data with plain-language notes. A machine translation that expands a compact abbreviation into an attractive sentence can lose the distinction between observation, plan, possibility and confirmation. The correct response is to resolve the underlying semantics, not to hide uncertainty behind polished target-language style.
Illustrative diagnostic. ‘Investigating suspected cable issue’ must not become ‘we confirmed cable failure’. The example is fictional and demonstrates an interpretation boundary rather than a live network condition. Translate within the assigned disclosure and evidence boundary. Preserve the original record in a review ledger so that a second bilingual reviewer can evaluate the decision without relying on the translator’s memory.
Release test. Check consistency between incident record, external advisory and update identifier. Read the target without the source and ask what action, belief or inference it encourages. Read both versions together and check whether the same asset, time, amount, conditionality and responsibility survive. If the two tests disagree, do not approve the rendering until the specific gap is resolved. Keep the technical judgement with the authorised domain specialist and the linguistic judgement with the accountable translator.
31. Cyber-physical event attribution
What must remain true. An unavailable control link is not proof of malicious activity. In professional cross-language work, the central unit of analysis is not the isolated word but the claim carried by an identified source. A term may have a direct bilingual equivalent while still needing a time qualifier, data type or institutional role to become accurate. The translator should name the controlled object first, then decide how a target reader is expected to use the sentence.
Why fluency is not enough. Security terms can overstate evidence when applied to an unexplained technical failure. This is especially important where a source mixes engineering data with plain-language notes. A machine translation that expands a compact abbreviation into an attractive sentence can lose the distinction between observation, plan, possibility and confirmation. The correct response is to resolve the underlying semantics, not to hide uncertainty behind polished target-language style.
Illustrative diagnostic. ‘Telemetry gateway unreachable’ is not equivalent to ‘cyberattack shut down the station’. The example is fictional and demonstrates an interpretation boundary rather than a live network condition. Preserve evidence category and escalation status; protect sensitive details without falsifying meaning. Preserve the original record in a review ledger so that a second bilingual reviewer can evaluate the decision without relying on the translator’s memory.
Release test. Use the approved incident classification and response communications owner. Read the target without the source and ask what action, belief or inference it encourages. Read both versions together and check whether the same asset, time, amount, conditionality and responsibility survive. If the two tests disagree, do not approve the rendering until the specific gap is resolved. Keep the technical judgement with the authorised domain specialist and the linguistic judgement with the accountable translator.
32. Audit records and regulatory evidence
What must remain true. The translation must preserve who asserted what, when and under which version. In professional cross-language work, the central unit of analysis is not the isolated word but the claim carried by an identified source. A term may have a direct bilingual equivalent while still needing a time qualifier, data type or institutional role to become accurate. The translator should name the controlled object first, then decide how a target reader is expected to use the sentence.
Why fluency is not enough. A later correction can be mistaken for the original event, and unsigned text for an authorised report. This is especially important where a source mixes engineering data with plain-language notes. A machine translation that expands a compact abbreviation into an attractive sentence can lose the distinction between observation, plan, possibility and confirmation. The correct response is to resolve the underlying semantics, not to hide uncertainty behind polished target-language style.
Illustrative diagnostic. ‘Draft incident finding’ is not a final determination. The example is fictional and demonstrates an interpretation boundary rather than a live network condition. Record source version, author, approval state, amendments and target-language revision history. Preserve the original record in a review ledger so that a second bilingual reviewer can evaluate the decision without relying on the translator’s memory.
Release test. Ask whether a third party can reproduce the claim chain without relying on an undocumented translator assumption. Read the target without the source and ask what action, belief or inference it encourages. Read both versions together and check whether the same asset, time, amount, conditionality and responsibility survive. If the two tests disagree, do not approve the rendering until the specific gap is resolved. Keep the technical judgement with the authorised domain specialist and the linguistic judgement with the accountable translator.
Mechanism synthesis: eight kinds of meaning that do not travel automatically
Identity is the first invariant. Codes, references, UUIDs, device names, event IDs and immutable message keys connect a language rendering to a source record. They should not be treated as a marketing opportunity. A label can have variants for different audiences, but a record key must remain referentially stable. Where the source identifier is uncertain, carry the uncertainty and ask the data owner.
State is the second invariant. A component can be open, closed, energized, unavailable, under maintenance, bypassed, tripped or unknown depending on the specific data model. Source-system state vocabularies may differ. Before choosing a translation, verify whether the term is a physical state, a supervisory indication, a workflow status or a human summary. These categories are related but not equivalent.
Time is the third invariant. Effective date, scheduled date, observed timestamp, event occurrence, last updated time, predicted restoration time and public notification time answer different questions. A language using aspect differently from the source may need extra words to keep that distinction. Never convert timezones without authority and without labelling the converted basis.
Quantity is the fourth invariant. Engineering expressions carry dimensions, reference conditions, tolerances, uncertainty, sign conventions and sometimes probabilistic ranges. Values printed in a table should usually remain traceable to source data; editorial prose must not produce a different value simply for easier reading. A consistent decimal style cannot repair an incorrectly typed unit.
Agency and permission form the fifth invariant. A relay may operate automatically; an operator may acknowledge an alarm; a crew may be assigned; a supervisor may authorise work; an application may calculate a suggested setpoint. A passive voice construction can conceal who acted, but the source does not necessarily authorise a translator to fill in that actor. Preserve known agents and leave unknown actors unknown.
Confidence forms the sixth invariant. Suspected faults, provisional topology, estimated customer counts and tentative restoration windows are not findings. The target must retain the strongest evidence-supported claim, not the strongest claim that sounds reassuring. This often means preserving words such as estimated, indicated, reported, under investigation, awaiting confirmation, or subject to change.
Audience is the seventh invariant. A professional operations record and a general customer notice may be about one event but have different safety, privacy and comprehension requirements. Technical density can be reduced for the customer, but not by changing claims. The writer can add a clearly marked explanation when authorised; the translator should not silently add a new engineering conclusion.
Revision and ownership are the eighth invariant. If a message is corrected, superseded or withdrawn, each language version needs to reflect that lifecycle. A reference to an old public notice may still be discoverable even when it is no longer authoritative. Translation should therefore carry a release ID, source version and a clear path to current information instead of depending on untracked copy-and-paste text.
The repeatable method: twenty-four gates for a translation release that can be audited
A reliable electricity-operations translation is produced through controlled decisions rather than a single fluent draft. These gates do not replace a utility’s safety, engineering, cybersecurity or change-management procedures; they organise the translator’s work so that specialists can approve their own domain. Every gate produces a testable artefact: a source snapshot, an identified asset, a verified quantity, a resolved ambiguity, a comparison record or a release decision. A gate that cannot be passed is recorded as an exception instead of being silently skipped.
Gate 01. Identify the receiving audience
Decision. Decide whether the content is intended for control-room personnel, engineers, field workers, public customers, regulators, researchers or customer-support agents. The reviewer should begin by identifying what would count as evidence that the decision is correct. A fluent rendering alone does not pass a gate; the preserved claim must also be fit for the intended reader and the source’s authority.
Action. Record channel, reading level, accessibility needs, language variety and whether the target is instructional, descriptive, evidential or promotional. Where the needed information does not exist, ask the source owner to clarify and preserve the unresolved state in the translation record. Translation is not a substitute for determining an engineering, regulatory or commercial fact. The source’s limits must travel into the target language.
Evidence to retain. An audience matrix linked to the source publication record. The document should be understandable by a second reviewer who did not participate in the first discussion. Failure this gate prevents: A single polished paragraph distributed to every reader despite material differences in what each is authorised to act on. Before closing the gate, test one realistic counterexample that would produce a different interpretation and show why the target does not permit it.
Gate 02. Confirm the source owner
Decision. Find the person or system with authority to explain the underlying record and approve its factual interpretation. The reviewer should begin by identifying what would count as evidence that the decision is correct. A fluent rendering alone does not pass a gate; the preserved claim must also be fit for the intended reader and the source’s authority.
Action. Mark the data originator, operations approver, public communications owner and translation reviewer separately; a translator must not assume those jobs. Where the needed information does not exist, ask the source owner to clarify and preserve the unresolved state in the translation record. Translation is not a substitute for determining an engineering, regulatory or commercial fact. The source’s limits must travel into the target language.
Evidence to retain. A responsibility map with named approval roles, not necessarily public names. The document should be understandable by a second reviewer who did not participate in the first discussion. Failure this gate prevents: A bilingual editor silently resolving a disputed equipment state without access to technical evidence. Before closing the gate, test one realistic counterexample that would produce a different interpretation and show why the target does not permit it.
Gate 03. Freeze the source version
Decision. Prevent a moving draft from producing multiple inconsistent language versions. The reviewer should begin by identifying what would count as evidence that the decision is correct. A fluent rendering alone does not pass a gate; the preserved claim must also be fit for the intended reader and the source’s authority.
Action. Capture the source version, time, content hash or stable export reference and record whether older customer notices must be superseded. Where the needed information does not exist, ask the source owner to clarify and preserve the unresolved state in the translation record. Translation is not a substitute for determining an engineering, regulatory or commercial fact. The source’s limits must travel into the target language.
Evidence to retain. An immutable reference to the source text used for the particular translation batch. The document should be understandable by a second reviewer who did not participate in the first discussion. Failure this gate prevents: One language variant being based on an estimate that has already been withdrawn from the source channel. Before closing the gate, test one realistic counterexample that would produce a different interpretation and show why the target does not permit it.
Gate 04. Identify the event and asset keys
Decision. Separate IDs and controlled labels from ordinary vocabulary. The reviewer should begin by identifying what would count as evidence that the decision is correct. A fluent rendering alone does not pass a gate; the preserved claim must also be fit for the intended reader and the source’s authority.
Action. Extract feeder IDs, equipment IDs, ticket numbers, timestamps, event codes, customer-message IDs and network-model keys into a protected field list. Where the needed information does not exist, ask the source owner to clarify and preserve the unresolved state in the translation record. Translation is not a substitute for determining an engineering, regulatory or commercial fact. The source’s limits must travel into the target language.
Evidence to retain. An identifier ledger with explicit copy-as-is or formally-mapped status. The document should be understandable by a second reviewer who did not participate in the first discussion. Failure this gate prevents: Localising a technical code because it resembles an everyday word and thereby losing cross-system traceability. Before closing the gate, test one realistic counterexample that would produce a different interpretation and show why the target does not permit it.
Gate 05. Classify every statement type
Decision. Distinguish observation, inference, planned action, issued instruction, completed action, estimate and public explanation. The reviewer should begin by identifying what would count as evidence that the decision is correct. A fluent rendering alone does not pass a gate; the preserved claim must also be fit for the intended reader and the source’s authority.
Action. Annotate consequential clauses by type, with special attention to verbs that can appear in either passive or active voice. Where the needed information does not exist, ask the source owner to clarify and preserve the unresolved state in the translation record. Translation is not a substitute for determining an engineering, regulatory or commercial fact. The source’s limits must travel into the target language.
Evidence to retain. A sentence-level claim map with the permitted evidential strength. The document should be understandable by a second reviewer who did not participate in the first discussion. Failure this gate prevents: Turning ‘command requested’ into ‘equipment operated’ or ‘suspected cause’ into ‘confirmed defect’. Before closing the gate, test one realistic counterexample that would produce a different interpretation and show why the target does not permit it.
Gate 06. Map electrical quantities
Decision. Ensure that values cannot change their physical dimensions through translation. The reviewer should begin by identifying what would count as evidence that the decision is correct. A fluent rendering alone does not pass a gate; the preserved claim must also be fit for the intended reader and the source’s authority.
Action. Tag V, kV, A, kA, kW, MW, kvar, Mvar, kVA, MVA, kWh, MWh, Hz, percentages, energy intervals and rating conditions as typed quantities. Where the needed information does not exist, ask the source owner to clarify and preserve the unresolved state in the translation record. Translation is not a substitute for determining an engineering, regulatory or commercial fact. The source’s limits must travel into the target language.
Evidence to retain. A value-unit register that can be checked mechanically. The document should be understandable by a second reviewer who did not participate in the first discussion. Failure this gate prevents: A 2.5 MW power claim becoming 2.5 MWh energy, or a calculated estimate becoming a meter reading. Before closing the gate, test one realistic counterexample that would produce a different interpretation and show why the target does not permit it.
Gate 07. Establish temporal meaning
Decision. Keep occurrence time, observation time, issue time and forecast effective time independent. The reviewer should begin by identifying what would count as evidence that the decision is correct. A fluent rendering alone does not pass a gate; the preserved claim must also be fit for the intended reader and the source’s authority.
Action. Record original timezone and offsets, reference the correct clock and preserve ‘as of’, ‘until’, ‘from’, ‘estimated’ and revision boundaries. Where the needed information does not exist, ask the source owner to clarify and preserve the unresolved state in the translation record. Translation is not a substitute for determining an engineering, regulatory or commercial fact. The source’s limits must travel into the target language.
Evidence to retain. A timeline of source events and publication changes. The document should be understandable by a second reviewer who did not participate in the first discussion. Failure this gate prevents: A later public notice suggesting a stale data point is current because its source timestamp disappeared. Before closing the gate, test one realistic counterexample that would produce a different interpretation and show why the target does not permit it.
Gate 08. Inspect network-model relationships
Decision. Ensure that connectedness, boundaries, ownership and topology stay intact. The reviewer should begin by identifying what would count as evidence that the decision is correct. A fluent rendering alone does not pass a gate; the preserved claim must also be fit for the intended reader and the source’s authority.
Action. Use an approved one-line diagram or data dictionary when a record depends on which feeder, busbar, transformer or section is connected to another. Where the needed information does not exist, ask the source owner to clarify and preserve the unresolved state in the translation record. Translation is not a substitute for determining an engineering, regulatory or commercial fact. The source’s limits must travel into the target language.
Evidence to retain. A concise asset-relationship map adequate for linguistic verification. The document should be understandable by a second reviewer who did not participate in the first discussion. Failure this gate prevents: Confusing a substation-side status with end-customer service, or attributing an event to the wrong network segment. Before closing the gate, test one realistic counterexample that would produce a different interpretation and show why the target does not permit it.
Gate 09. Check authority and permission language
Decision. Separate human authorization from a description of equipment or workflow status. The reviewer should begin by identifying what would count as evidence that the decision is correct. A fluent rendering alone does not pass a gate; the preserved claim must also be fit for the intended reader and the source’s authority.
Action. Highlight may, must, should, shall, approved, requested, blocked, permitted, forbidden and pending, then verify institutional meanings. Where the needed information does not exist, ask the source owner to clarify and preserve the unresolved state in the translation record. Translation is not a substitute for determining an engineering, regulatory or commercial fact. The source’s limits must travel into the target language.
Evidence to retain. A modality and approval-state table. The document should be understandable by a second reviewer who did not participate in the first discussion. Failure this gate prevents: An unapproved request being presented as a valid instruction or a safety clearance. Before closing the gate, test one realistic counterexample that would produce a different interpretation and show why the target does not permit it.
Gate 10. Prepare a concept-based termbase
Decision. Use definitions and relations rather than one-to-one bilingual guesses. The reviewer should begin by identifying what would count as evidence that the decision is correct. A fluent rendering alone does not pass a gate; the preserved claim must also be fit for the intended reader and the source’s authority.
Action. Include source term, definition, approved target term, forbidden variants, domain, audience, standard version, example and decision owner. Where the needed information does not exist, ask the source owner to clarify and preserve the unresolved state in the translation record. Translation is not a substitute for determining an engineering, regulatory or commercial fact. The source’s limits must travel into the target language.
Evidence to retain. A revision-controlled bilingual termbase with concept IDs. The document should be understandable by a second reviewer who did not participate in the first discussion. Failure this gate prevents: Applying one target term to breaker, disconnector and protective relay simply because ordinary usage calls each a switch. Before closing the gate, test one realistic counterexample that would produce a different interpretation and show why the target does not permit it.
Gate 11. Record language-pair contrasts
Decision. Find grammatical features that can distort agency, conditionality or time. The reviewer should begin by identifying what would count as evidence that the decision is correct. A fluent rendering alone does not pass a gate; the preserved claim must also be fit for the intended reader and the source’s authority.
Action. Check whether the target language requires more explicit subjects, aspect, number, evidential markers, classifiers or politeness distinctions. Where the needed information does not exist, ask the source owner to clarify and preserve the unresolved state in the translation record. Translation is not a substitute for determining an engineering, regulatory or commercial fact. The source’s limits must travel into the target language.
Evidence to retain. A brief language-pair risk sheet with examples from this actual project. The document should be understandable by a second reviewer who did not participate in the first discussion. Failure this gate prevents: A target sentence asserting completion because a source progressive or conditional construction was collapsed. Before closing the gate, test one realistic counterexample that would produce a different interpretation and show why the target does not permit it.
Gate 12. Control names, scripts and transliteration
Decision. Preserve identity across writing systems and avoid multiple competing spellings. The reviewer should begin by identifying what would count as evidence that the decision is correct. A fluent rendering alone does not pass a gate; the preserved claim must also be fit for the intended reader and the source’s authority.
Action. Decide which labels may be translated, which personal and organisation names have official forms, and which equipment codes must remain unchanged. Where the needed information does not exist, ask the source owner to clarify and preserve the unresolved state in the translation record. Translation is not a substitute for determining an engineering, regulatory or commercial fact. The source’s limits must travel into the target language.
Evidence to retain. A name and label register tied to source entities. The document should be understandable by a second reviewer who did not participate in the first discussion. Failure this gate prevents: A translated customer notice that sounds familiar but names the wrong site or work party. Before closing the gate, test one realistic counterexample that would produce a different interpretation and show why the target does not permit it.
Gate 13. Separate structured data and display text
Decision. Prevent localisation from corrupting an interoperable exchange payload. The reviewer should begin by identifying what would count as evidence that the decision is correct. A fluent rendering alone does not pass a gate; the preserved claim must also be fit for the intended reader and the source’s authority.
Action. Identify schema keys, enum values, machine event codes, human descriptions and multilingual display fields before touching any string. Where the needed information does not exist, ask the source owner to clarify and preserve the unresolved state in the translation record. Translation is not a substitute for determining an engineering, regulatory or commercial fact. The source’s limits must travel into the target language.
Evidence to retain. A protected-fields inventory and test data fixture. The document should be understandable by a second reviewer who did not participate in the first discussion. Failure this gate prevents: Translating a Common Information Model class name or IEC 61850 data-object identifier inside a production exchange. Before closing the gate, test one realistic counterexample that would produce a different interpretation and show why the target does not permit it.
Gate 14. Draft semantic units
Decision. Translate claims, conditions and referents together rather than isolated strings. The reviewer should begin by identifying what would count as evidence that the decision is correct. A fluent rendering alone does not pass a gate; the preserved claim must also be fit for the intended reader and the source’s authority.
Action. Keep each action attached to its actor, equipment, relevant time and constraints; allow natural target syntax while preserving these relations. Where the needed information does not exist, ask the source owner to clarify and preserve the unresolved state in the translation record. Translation is not a substitute for determining an engineering, regulatory or commercial fact. The source’s limits must travel into the target language.
Evidence to retain. A coherent first-pass target that exposes its source-to-target mapping. The document should be understandable by a second reviewer who did not participate in the first discussion. Failure this gate prevents: A tiny interface label that omits a necessary condition because the string was extracted without surrounding context. Before closing the gate, test one realistic counterexample that would produce a different interpretation and show why the target does not permit it.
Gate 15. Compare source and target assertions
Decision. Test whether the target asserts more, less or something different from the original. The reviewer should begin by identifying what would count as evidence that the decision is correct. A fluent rendering alone does not pass a gate; the preserved claim must also be fit for the intended reader and the source’s authority.
Action. Use a bilingual comparison that asks which entity, state, unit, cause, time, certainty and authorized outcome a reader believes. Where the needed information does not exist, ask the source owner to clarify and preserve the unresolved state in the translation record. Translation is not a substitute for determining an engineering, regulatory or commercial fact. The source’s limits must travel into the target language.
Evidence to retain. An assertion-difference table with unresolved issues marked. The document should be understandable by a second reviewer who did not participate in the first discussion. Failure this gate prevents: A grammatically perfect version that promises full restoration when only partial restoration was reported. Before closing the gate, test one realistic counterexample that would produce a different interpretation and show why the target does not permit it.
Gate 16. Review safety-critical meaning
Decision. Escalate phrases that could influence hazardous real-world decisions. The reviewer should begin by identifying what would count as evidence that the decision is correct. A fluent rendering alone does not pass a gate; the preserved claim must also be fit for the intended reader and the source’s authority.
Action. Have a competent authorized engineering or operations reviewer evaluate the source-bound meanings of states, prohibitions, warnings and approvals. Where the needed information does not exist, ask the source owner to clarify and preserve the unresolved state in the translation record. Translation is not a substitute for determining an engineering, regulatory or commercial fact. The source’s limits must travel into the target language.
Evidence to retain. A safety-meaning review record with reviewer and version. The document should be understandable by a second reviewer who did not participate in the first discussion. Failure this gate prevents: The translator providing an improvised switching sequence, permission or protective recommendation. Before closing the gate, test one realistic counterexample that would produce a different interpretation and show why the target does not permit it.
Gate 17. Validate numeric invariants
Decision. Catch invisible corruption of values, symbols and delimiters. The reviewer should begin by identifying what would count as evidence that the decision is correct. A fluent rendering alone does not pass a gate; the preserved claim must also be fit for the intended reader and the source’s authority.
Action. Compare source and target numeric fields, signs, inequalities, interval endpoints, unit prefixes and decimal/thousands conventions using a typed check. Where the needed information does not exist, ask the source owner to clarify and preserve the unresolved state in the translation record. Translation is not a substitute for determining an engineering, regulatory or commercial fact. The source’s limits must travel into the target language.
Evidence to retain. A numeric parity report and evidence of any approved conversion. The document should be understandable by a second reviewer who did not participate in the first discussion. Failure this gate prevents: A minus sign disappearing from a net-export quantity or an upper bound becoming a lower bound. Before closing the gate, test one realistic counterexample that would produce a different interpretation and show why the target does not permit it.
Gate 18. Validate time and schedule rendering
Decision. Prevent customers or operators from acting on the wrong effective period. The reviewer should begin by identifying what would count as evidence that the decision is correct. A fluent rendering alone does not pass a gate; the preserved claim must also be fit for the intended reader and the source’s authority.
Action. Test timezone, daylight-saving assumptions, calendar dates, inclusive and exclusive intervals, frequency of updates and provisional forecasts. Where the needed information does not exist, ask the source owner to clarify and preserve the unresolved state in the translation record. Translation is not a substitute for determining an engineering, regulatory or commercial fact. The source’s limits must travel into the target language.
Evidence to retain. A time-parity table and displayed timezone label. The document should be understandable by a second reviewer who did not participate in the first discussion. Failure this gate prevents: An estimated 18:00 UTC being published as 18:00 local without disclosure. Before closing the gate, test one realistic counterexample that would produce a different interpretation and show why the target does not permit it.
Gate 19. Validate system interoperability
Decision. Check that machine-readable messages still parse and maintain identity. The reviewer should begin by identifying what would count as evidence that the decision is correct. A fluent rendering alone does not pass a gate; the preserved claim must also be fit for the intended reader and the source’s authority.
Action. Run the relevant engineering-approved schema and integration tests; translation owns explanatory text and cannot infer protocol compliance. Where the needed information does not exist, ask the source owner to clarify and preserve the unresolved state in the translation record. Translation is not a substitute for determining an engineering, regulatory or commercial fact. The source’s limits must travel into the target language.
Evidence to retain. A comparison of object counts, keys, enums and relations before and after localisation. The document should be understandable by a second reviewer who did not participate in the first discussion. Failure this gate prevents: A screen looking correct while the exchanged data silently changes class, code or reference. Before closing the gate, test one realistic counterexample that would produce a different interpretation and show why the target does not permit it.
Gate 20. Test target-language comprehension
Decision. Confirm that a reader does not infer a new operational promise. The reviewer should begin by identifying what would count as evidence that the decision is correct. A fluent rendering alone does not pass a gate; the preserved claim must also be fit for the intended reader and the source’s authority.
Action. Ask representative target readers to explain what is known, not known, scheduled, estimated and required, then compare with intended meaning. Where the needed information does not exist, ask the source owner to clarify and preserve the unresolved state in the translation record. Translation is not a substitute for determining an engineering, regulatory or commercial fact. The source’s limits must travel into the target language.
Evidence to retain. A comprehension test showing questions, answers and required fixes. The document should be understandable by a second reviewer who did not participate in the first discussion. Failure this gate prevents: Readers believing that ‘crew assigned’ means ‘power will return soon’ after a vague rewrite. Before closing the gate, test one realistic counterexample that would produce a different interpretation and show why the target does not permit it.
Gate 21. Test accessibility and constrained channels
Decision. Ensure short messages, screen readers and mobile notices retain their critical qualifications. The reviewer should begin by identifying what would count as evidence that the decision is correct. A fluent rendering alone does not pass a gate; the preserved claim must also be fit for the intended reader and the source’s authority.
Action. Evaluate line wrapping, abbreviation definitions, alternative formats, Unicode symbols, bidirectional text and channels that truncate words. Where the needed information does not exist, ask the source owner to clarify and preserve the unresolved state in the translation record. Translation is not a substitute for determining an engineering, regulatory or commercial fact. The source’s limits must travel into the target language.
Evidence to retain. Screenshots or render tests with key information still discoverable. The document should be understandable by a second reviewer who did not participate in the first discussion. Failure this gate prevents: An SMS losing the word estimated while keeping a confident time and a call-to-action. Before closing the gate, test one realistic counterexample that would produce a different interpretation and show why the target does not permit it.
Gate 22. Maintain an exception ledger
Decision. Make uncertainty visible without allowing a hidden guessed translation to go live. The reviewer should begin by identifying what would count as evidence that the decision is correct. A fluent rendering alone does not pass a gate; the preserved claim must also be fit for the intended reader and the source’s authority.
Action. Log source ambiguity, missing evidence, unresolved term conflicts and policy questions; assign owners and deadlines, and block risky release items. Where the needed information does not exist, ask the source owner to clarify and preserve the unresolved state in the translation record. Translation is not a substitute for determining an engineering, regulatory or commercial fact. The source’s limits must travel into the target language.
Evidence to retain. An exception record containing decision, authority, rationale and affected languages. The document should be understandable by a second reviewer who did not participate in the first discussion. Failure this gate prevents: The source being corrected later with no way to identify which language versions inherited the original problem. Before closing the gate, test one realistic counterexample that would produce a different interpretation and show why the target does not permit it.
Gate 23. Approve and publish together
Decision. Prevent source and target language channels from drifting during a live event. The reviewer should begin by identifying what would count as evidence that the decision is correct. A fluent rendering alone does not pass a gate; the preserved claim must also be fit for the intended reader and the source’s authority.
Action. Confirm which version is current, which older notices are superseded, and who authorizes release for each audience and channel. Where the needed information does not exist, ask the source owner to clarify and preserve the unresolved state in the translation record. Translation is not a substitute for determining an engineering, regulatory or commercial fact. The source’s limits must travel into the target language.
Evidence to retain. A versioned publication manifest and a clear rollback or withdrawal route. The document should be understandable by a second reviewer who did not participate in the first discussion. Failure this gate prevents: A translated alert remaining visible after the source has been replaced with an all-clear. Before closing the gate, test one realistic counterexample that would produce a different interpretation and show why the target does not permit it.
Gate 24. Monitor, learn and revise
Decision. Treat translation quality as a repeated operational process, not a one-time act. The reviewer should begin by identifying what would count as evidence that the decision is correct. A fluent rendering alone does not pass a gate; the preserved claim must also be fit for the intended reader and the source’s authority.
Action. Review complaints, corrections, repeated term failures and false inferences; update the termbase and tests rather than only patching one sentence. Where the needed information does not exist, ask the source owner to clarify and preserve the unresolved state in the translation record. Translation is not a substitute for determining an engineering, regulatory or commercial fact. The source’s limits must travel into the target language.
Evidence to retain. A feedback log and revised controls that prevent recurrence. The document should be understandable by a second reviewer who did not participate in the first discussion. Failure this gate prevents: Repeatedly discovering the same distinction between estimated and confirmed restoration in different projects. Before closing the gate, test one realistic counterexample that would produce a different interpretation and show why the target does not permit it.
Failure modes: forty ways fluent utility translation goes wrong
Failures are easier to prevent when each has a recognizable signature. The examples below are diagnostic categories, not evidence about any actual operator or network. A correct review must still consider the approved data model, source version, audience and language pair. Where the source is itself wrong or uncertain, record that fact rather than quietly laundering it into certainty through translation.
Asset drift
Signature. An equipment code is adapted as ordinary prose. Why it matters. The target no longer resolves to the original equipment record. The mistake may be invisible to a monolingual reader and can pass a superficial fluency review. The failure is semantic: the target version leads the audience to believe a different proposition about an electrical system, an action or an uncertainty.
Repair. Protect keys and review the source-to-target entity map. Compare the disputed statement with its original source field and relevant neighbouring records. Ask whether the reader can still recover the correct actor, time, asset, value, status and authority. If evidence is incomplete, preserve the incompleteness and refer the decision to the responsible specialist rather than creating a confident substitute.
Object-type substitution
Signature. Breaker, disconnector, fuse and relay are given one generic target word. Why it matters. The translated document may imply a different protection or switching function. The mistake may be invisible to a monolingual reader and can pass a superficial fluency review. The failure is semantic: the target version leads the audience to believe a different proposition about an electrical system, an action or an uncertainty.
Repair. Use equipment classes from the controlled technical dictionary. Compare the disputed statement with its original source field and relevant neighbouring records. Ask whether the reader can still recover the correct actor, time, asset, value, status and authority. If evidence is incomplete, preserve the incompleteness and refer the decision to the responsible specialist rather than creating a confident substitute.
Topology inflation
Signature. A connected network diagram is described as confirmed energized supply. Why it matters. The claim leaps from a model relationship to a real-world service state. The mistake may be invisible to a monolingual reader and can pass a superficial fluency review. The failure is semantic: the target version leads the audience to believe a different proposition about an electrical system, an action or an uncertainty.
Repair. Distinguish modeled topology from measured or verified conditions. Compare the disputed statement with its original source field and relevant neighbouring records. Ask whether the reader can still recover the correct actor, time, asset, value, status and authority. If evidence is incomplete, preserve the incompleteness and refer the decision to the responsible specialist rather than creating a confident substitute.
Command-versus-feedback collapse
Signature. A queued request is reported as an accomplished act. Why it matters. A future or pending operation becomes a historical assertion. The mistake may be invisible to a monolingual reader and can pass a superficial fluency review. The failure is semantic: the target version leads the audience to believe a different proposition about an electrical system, an action or an uncertainty.
Repair. Match command, acknowledgement and feedback as separate events. Compare the disputed statement with its original source field and relevant neighbouring records. Ask whether the reader can still recover the correct actor, time, asset, value, status and authority. If evidence is incomplete, preserve the incompleteness and refer the decision to the responsible specialist rather than creating a confident substitute.
Protective-action misattribution
Signature. A protection trip becomes a human-initiated operation. Why it matters. The message changes both agent and cause. The mistake may be invisible to a monolingual reader and can pass a superficial fluency review. The failure is semantic: the target version leads the audience to believe a different proposition about an electrical system, an action or an uncertainty.
Repair. Keep automatic protection and manual control vocabulary distinct. Compare the disputed statement with its original source field and relevant neighbouring records. Ask whether the reader can still recover the correct actor, time, asset, value, status and authority. If evidence is incomplete, preserve the incompleteness and refer the decision to the responsible specialist rather than creating a confident substitute.
Implied work permission
Signature. De-energized wording is recast as safe-to-work authorization. Why it matters. The target may imply permissions that the source never granted. The mistake may be invisible to a monolingual reader and can pass a superficial fluency review. The failure is semantic: the target version leads the audience to believe a different proposition about an electrical system, an action or an uncertainty.
Repair. Escalate to the authorized safety reviewer and preserve the limited claim. Compare the disputed statement with its original source field and relevant neighbouring records. Ask whether the reader can still recover the correct actor, time, asset, value, status and authority. If evidence is incomplete, preserve the incompleteness and refer the decision to the responsible specialist rather than creating a confident substitute.
Source-cause invention
Signature. A relay alarm is translated as a confirmed cable failure. Why it matters. A symptom is promoted to diagnosis. The mistake may be invisible to a monolingual reader and can pass a superficial fluency review. The failure is semantic: the target version leads the audience to believe a different proposition about an electrical system, an action or an uncertainty.
Repair. Retain suspected, indicated and confirmed classifications. Compare the disputed statement with its original source field and relevant neighbouring records. Ask whether the reader can still recover the correct actor, time, asset, value, status and authority. If evidence is incomplete, preserve the incompleteness and refer the decision to the responsible specialist rather than creating a confident substitute.
Status aging
Signature. A previous valid value appears as a current condition. Why it matters. Readers cannot tell when the observation was obtained. The mistake may be invisible to a monolingual reader and can pass a superficial fluency review. The failure is semantic: the target version leads the audience to believe a different proposition about an electrical system, an action or an uncertainty.
Repair. Display sampling time and freshness/quality when relevant. Compare the disputed statement with its original source field and relevant neighbouring records. Ask whether the reader can still recover the correct actor, time, asset, value, status and authority. If evidence is incomplete, preserve the incompleteness and refer the decision to the responsible specialist rather than creating a confident substitute.
Erased data quality
Signature. Invalid or unknown flags disappear from a translated dashboard. Why it matters. A value appears more reliable than the evidence permits. The mistake may be invisible to a monolingual reader and can pass a superficial fluency review. The failure is semantic: the target version leads the audience to believe a different proposition about an electrical system, an action or an uncertainty.
Repair. Retain quality flags and uncertainty indicators. Compare the disputed statement with its original source field and relevant neighbouring records. Ask whether the reader can still recover the correct actor, time, asset, value, status and authority. If evidence is incomplete, preserve the incompleteness and refer the decision to the responsible specialist rather than creating a confident substitute.
MW/MWh confusion
Signature. A rate of power is described as accumulated energy. Why it matters. The target changes the dimensional meaning of the record. The mistake may be invisible to a monolingual reader and can pass a superficial fluency review. The failure is semantic: the target version leads the audience to believe a different proposition about an electrical system, an action or an uncertainty.
Repair. Check type, unit and time interval. Compare the disputed statement with its original source field and relevant neighbouring records. Ask whether the reader can still recover the correct actor, time, asset, value, status and authority. If evidence is incomplete, preserve the incompleteness and refer the decision to the responsible specialist rather than creating a confident substitute.
MVA/MW confusion
Signature. Apparent and active power are translated as synonyms. Why it matters. Capacity and actual active output are falsely aligned. The mistake may be invisible to a monolingual reader and can pass a superficial fluency review. The failure is semantic: the target version leads the audience to believe a different proposition about an electrical system, an action or an uncertainty.
Repair. Preserve dimensional distinction and any power-factor context. Compare the disputed statement with its original source field and relevant neighbouring records. Ask whether the reader can still recover the correct actor, time, asset, value, status and authority. If evidence is incomplete, preserve the incompleteness and refer the decision to the responsible specialist rather than creating a confident substitute.
Current/voltage mix-up
Signature. A numerical reading is attached to the wrong electrical quantity. Why it matters. Even correct digits become misleading. The mistake may be invisible to a monolingual reader and can pass a superficial fluency review. The failure is semantic: the target version leads the audience to believe a different proposition about an electrical system, an action or an uncertainty.
Repair. Bind value with quantity symbol, unit and measurement point. Compare the disputed statement with its original source field and relevant neighbouring records. Ask whether the reader can still recover the correct actor, time, asset, value, status and authority. If evidence is incomplete, preserve the incompleteness and refer the decision to the responsible specialist rather than creating a confident substitute.
Prefix shift
Signature. kW becomes MW or kV becomes V without consistent scaling. Why it matters. A magnitude changes by orders of magnitude. The mistake may be invisible to a monolingual reader and can pass a superficial fluency review. The failure is semantic: the target version leads the audience to believe a different proposition about an electrical system, an action or an uncertainty.
Repair. Run a typed quantity and unit-prefix audit. Compare the disputed statement with its original source field and relevant neighbouring records. Ask whether the reader can still recover the correct actor, time, asset, value, status and authority. If evidence is incomplete, preserve the incompleteness and refer the decision to the responsible specialist rather than creating a confident substitute.
Rounding drift
Signature. A 10.04 value becomes 10.4 through decimal-style mistakes. Why it matters. The source result changes while the prose looks normal. The mistake may be invisible to a monolingual reader and can pass a superficial fluency review. The failure is semantic: the target version leads the audience to believe a different proposition about an electrical system, an action or an uncertainty.
Repair. Validate locale numerals and significant digits. Compare the disputed statement with its original source field and relevant neighbouring records. Ask whether the reader can still recover the correct actor, time, asset, value, status and authority. If evidence is incomplete, preserve the incompleteness and refer the decision to the responsible specialist rather than creating a confident substitute.
Sign reversal
Signature. Import and export quantities lose their sign. Why it matters. Direction of energy exchange reverses. The mistake may be invisible to a monolingual reader and can pass a superficial fluency review. The failure is semantic: the target version leads the audience to believe a different proposition about an electrical system, an action or an uncertainty.
Repair. Preserve the network’s documented sign convention. Compare the disputed statement with its original source field and relevant neighbouring records. Ask whether the reader can still recover the correct actor, time, asset, value, status and authority. If evidence is incomplete, preserve the incompleteness and refer the decision to the responsible specialist rather than creating a confident substitute.
Threshold inversion
Signature. ‘Above’ becomes ‘at least’ or ‘below’ becomes ‘not above’. Why it matters. The set of conditions governed by the claim changes. The mistake may be invisible to a monolingual reader and can pass a superficial fluency review. The failure is semantic: the target version leads the audience to believe a different proposition about an electrical system, an action or an uncertainty.
Repair. Test boundary values on both sides of the threshold. Compare the disputed statement with its original source field and relevant neighbouring records. Ask whether the reader can still recover the correct actor, time, asset, value, status and authority. If evidence is incomplete, preserve the incompleteness and refer the decision to the responsible specialist rather than creating a confident substitute.
Timezone disappearance
Signature. UTC values are delivered as unlabeled clock times. Why it matters. The audience may act at the wrong local hour. The mistake may be invisible to a monolingual reader and can pass a superficial fluency review. The failure is semantic: the target version leads the audience to believe a different proposition about an electrical system, an action or an uncertainty.
Repair. Keep timezone explicit and document authorized conversion. Compare the disputed statement with its original source field and relevant neighbouring records. Ask whether the reader can still recover the correct actor, time, asset, value, status and authority. If evidence is incomplete, preserve the incompleteness and refer the decision to the responsible specialist rather than creating a confident substitute.
Forecast certainty
Signature. Estimated restoration becomes a fixed deadline. Why it matters. A planning forecast becomes a service guarantee. The mistake may be invisible to a monolingual reader and can pass a superficial fluency review. The failure is semantic: the target version leads the audience to believe a different proposition about an electrical system, an action or an uncertainty.
Repair. Maintain probabilistic or conditional language. Compare the disputed statement with its original source field and relevant neighbouring records. Ask whether the reader can still recover the correct actor, time, asset, value, status and authority. If evidence is incomplete, preserve the incompleteness and refer the decision to the responsible specialist rather than creating a confident substitute.
Publication-date conflation
Signature. The issue time is mistaken for the effective time. Why it matters. The target suggests an event has already begun or remains active. The mistake may be invisible to a monolingual reader and can pass a superficial fluency review. The failure is semantic: the target version leads the audience to believe a different proposition about an electrical system, an action or an uncertainty.
Repair. Compare event, effective, issue and update timestamps. Compare the disputed statement with its original source field and relevant neighbouring records. Ask whether the reader can still recover the correct actor, time, asset, value, status and authority. If evidence is incomplete, preserve the incompleteness and refer the decision to the responsible specialist rather than creating a confident substitute.
Partial/full collapse
Signature. Some restored customers are reported as all restored. Why it matters. Remaining affected groups disappear. The mistake may be invisible to a monolingual reader and can pass a superficial fluency review. The failure is semantic: the target version leads the audience to believe a different proposition about an electrical system, an action or an uncertainty.
Repair. Keep scope, exceptions and denominators explicit. Compare the disputed statement with its original source field and relevant neighbouring records. Ask whether the reader can still recover the correct actor, time, asset, value, status and authority. If evidence is incomplete, preserve the incompleteness and refer the decision to the responsible specialist rather than creating a confident substitute.
Geographical false precision
Signature. A suspected affected area becomes a precise street claim. Why it matters. Location confidence is exaggerated. The mistake may be invisible to a monolingual reader and can pass a superficial fluency review. The failure is semantic: the target version leads the audience to believe a different proposition about an electrical system, an action or an uncertainty.
Repair. Record source geometry and uncertainty. Compare the disputed statement with its original source field and relevant neighbouring records. Ask whether the reader can still recover the correct actor, time, asset, value, status and authority. If evidence is incomplete, preserve the incompleteness and refer the decision to the responsible specialist rather than creating a confident substitute.
Customer-count confusion
Signature. Calls received are presented as affected customers. Why it matters. A proxy becomes an authoritative total. The mistake may be invisible to a monolingual reader and can pass a superficial fluency review. The failure is semantic: the target version leads the audience to believe a different proposition about an electrical system, an action or an uncertainty.
Repair. Preserve whether the figure is reports, meters, accounts or premises. Compare the disputed statement with its original source field and relevant neighbouring records. Ask whether the reader can still recover the correct actor, time, asset, value, status and authority. If evidence is incomplete, preserve the incompleteness and refer the decision to the responsible specialist rather than creating a confident substitute.
Crew-assignment inflation
Signature. Dispatched or assigned personnel become confirmed onsite. Why it matters. A workflow stage appears completed. The mistake may be invisible to a monolingual reader and can pass a superficial fluency review. The failure is semantic: the target version leads the audience to believe a different proposition about an electrical system, an action or an uncertainty.
Repair. Use controlled verbs for each field-work stage. Compare the disputed statement with its original source field and relevant neighbouring records. Ask whether the reader can still recover the correct actor, time, asset, value, status and authority. If evidence is incomplete, preserve the incompleteness and refer the decision to the responsible specialist rather than creating a confident substitute.
Repair-versus-supply collapse
Signature. Equipment repair completion becomes restored electricity. Why it matters. A component milestone is mistaken for the customer outcome. The mistake may be invisible to a monolingual reader and can pass a superficial fluency review. The failure is semantic: the target version leads the audience to believe a different proposition about an electrical system, an action or an uncertainty.
Repair. Require explicit service-restoration evidence. Compare the disputed statement with its original source field and relevant neighbouring records. Ask whether the reader can still recover the correct actor, time, asset, value, status and authority. If evidence is incomplete, preserve the incompleteness and refer the decision to the responsible specialist rather than creating a confident substitute.
Planned/unplanned misclassification
Signature. Pending maintenance becomes a confirmed outage. Why it matters. Public expectations and timing change. The mistake may be invisible to a monolingual reader and can pass a superficial fluency review. The failure is semantic: the target version leads the audience to believe a different proposition about an electrical system, an action or an uncertainty.
Repair. Preserve work authorization and schedule status. Compare the disputed statement with its original source field and relevant neighbouring records. Ask whether the reader can still recover the correct actor, time, asset, value, status and authority. If evidence is incomplete, preserve the incompleteness and refer the decision to the responsible specialist rather than creating a confident substitute.
CIM schema translation
Signature. Technical class or property names are rewritten as display labels. Why it matters. Integration, data mapping and system provenance may break. The mistake may be invisible to a monolingual reader and can pass a superficial fluency review. The failure is semantic: the target version leads the audience to believe a different proposition about an electrical system, an action or an uncertainty.
Repair. Separate immutable schema fields from translated descriptions. Compare the disputed statement with its original source field and relevant neighbouring records. Ask whether the reader can still recover the correct actor, time, asset, value, status and authority. If evidence is incomplete, preserve the incompleteness and refer the decision to the responsible specialist rather than creating a confident substitute.
IEC 61850 label localization
Signature. Logical-node identifiers are localized as prose. Why it matters. Machine interfaces can cease to match a model. The mistake may be invisible to a monolingual reader and can pass a superficial fluency review. The failure is semantic: the target version leads the audience to believe a different proposition about an electrical system, an action or an uncertainty.
Repair. Keep protected codes stable and test profiles. Compare the disputed statement with its original source field and relevant neighbouring records. Ask whether the reader can still recover the correct actor, time, asset, value, status and authority. If evidence is incomplete, preserve the incompleteness and refer the decision to the responsible specialist rather than creating a confident substitute.
Meter-interval omission
Signature. A kWh value loses interval start and end. Why it matters. Consumption is detached from its measuring window. The mistake may be invisible to a monolingual reader and can pass a superficial fluency review. The failure is semantic: the target version leads the audience to believe a different proposition about an electrical system, an action or an uncertainty.
Repair. Preserve period and register/interval type. Compare the disputed statement with its original source field and relevant neighbouring records. Ask whether the reader can still recover the correct actor, time, asset, value, status and authority. If evidence is incomplete, preserve the incompleteness and refer the decision to the responsible specialist rather than creating a confident substitute.
Storage rating confusion
Signature. Battery MWh energy is presented as MW power. Why it matters. A capacity statement becomes an impossible dispatch claim. The mistake may be invisible to a monolingual reader and can pass a superficial fluency review. The failure is semantic: the target version leads the audience to believe a different proposition about an electrical system, an action or an uncertainty.
Repair. Keep power and stored-energy ratings distinct. Compare the disputed statement with its original source field and relevant neighbouring records. Ask whether the reader can still recover the correct actor, time, asset, value, status and authority. If evidence is incomplete, preserve the incompleteness and refer the decision to the responsible specialist rather than creating a confident substitute.
Renewables guarantee
Signature. Forecast solar output becomes guaranteed production. Why it matters. Uncertainty is hidden and planning intent changes. The mistake may be invisible to a monolingual reader and can pass a superficial fluency review. The failure is semantic: the target version leads the audience to believe a different proposition about an electrical system, an action or an uncertainty.
Repair. Retain forecast horizon and ranges. Compare the disputed statement with its original source field and relevant neighbouring records. Ask whether the reader can still recover the correct actor, time, asset, value, status and authority. If evidence is incomplete, preserve the incompleteness and refer the decision to the responsible specialist rather than creating a confident substitute.
Curtailment subject swap
Signature. Generation curtailment becomes customer demand reduction. Why it matters. The affected party and intervention differ. The mistake may be invisible to a monolingual reader and can pass a superficial fluency review. The failure is semantic: the target version leads the audience to believe a different proposition about an electrical system, an action or an uncertainty.
Repair. Name the source of energy and the direction of the action. Compare the disputed statement with its original source field and relevant neighbouring records. Ask whether the reader can still recover the correct actor, time, asset, value, status and authority. If evidence is incomplete, preserve the incompleteness and refer the decision to the responsible specialist rather than creating a confident substitute.
Demand-response consent loss
Signature. Optional participation becomes compulsory. Why it matters. A translated message changes a user’s choice. The mistake may be invisible to a monolingual reader and can pass a superficial fluency review. The failure is semantic: the target version leads the audience to believe a different proposition about an electrical system, an action or an uncertainty.
Repair. Preserve opt-in, eligibility and enrollment status. Compare the disputed statement with its original source field and relevant neighbouring records. Ask whether the reader can still recover the correct actor, time, asset, value, status and authority. If evidence is incomplete, preserve the incompleteness and refer the decision to the responsible specialist rather than creating a confident substitute.
Security conclusion leap
Signature. Telemetry outage becomes proof of malicious activity. Why it matters. A technical symptom becomes a sensitive accusation. The mistake may be invisible to a monolingual reader and can pass a superficial fluency review. The failure is semantic: the target version leads the audience to believe a different proposition about an electrical system, an action or an uncertainty.
Repair. Separate availability, investigation and confirmed classification. Compare the disputed statement with its original source field and relevant neighbouring records. Ask whether the reader can still recover the correct actor, time, asset, value, status and authority. If evidence is incomplete, preserve the incompleteness and refer the decision to the responsible specialist rather than creating a confident substitute.
Fault investigation chronology
Signature. A later finding is presented as known at the incident start. Why it matters. Historical claims are rewritten without disclosure. The mistake may be invisible to a monolingual reader and can pass a superficial fluency review. The failure is semantic: the target version leads the audience to believe a different proposition about an electrical system, an action or an uncertainty.
Repair. Maintain event timeline and revision metadata. Compare the disputed statement with its original source field and relevant neighbouring records. Ask whether the reader can still recover the correct actor, time, asset, value, status and authority. If evidence is incomplete, preserve the incompleteness and refer the decision to the responsible specialist rather than creating a confident substitute.
Omitted supersession
Signature. An older estimate remains visible after a correction. Why it matters. Different audiences receive contradictory versions. The mistake may be invisible to a monolingual reader and can pass a superficial fluency review. The failure is semantic: the target version leads the audience to believe a different proposition about an electrical system, an action or an uncertainty.
Repair. Tie language variants to revision and withdrawal controls. Compare the disputed statement with its original source field and relevant neighbouring records. Ask whether the reader can still recover the correct actor, time, asset, value, status and authority. If evidence is incomplete, preserve the incompleteness and refer the decision to the responsible specialist rather than creating a confident substitute.
Channel truncation
Signature. A short banner drops ‘estimated’ or ‘not yet confirmed’. Why it matters. Critical meaning is removed by interface length. The mistake may be invisible to a monolingual reader and can pass a superficial fluency review. The failure is semantic: the target version leads the audience to believe a different proposition about an electrical system, an action or an uncertainty.
Repair. Test actual rendered content across channels. Compare the disputed statement with its original source field and relevant neighbouring records. Ask whether the reader can still recover the correct actor, time, asset, value, status and authority. If evidence is incomplete, preserve the incompleteness and refer the decision to the responsible specialist rather than creating a confident substitute.
Machine-translation hallucination
Signature. The target invents an unmentioned crew, component or reason. Why it matters. Additional seemingly useful details are fabricated. The mistake may be invisible to a monolingual reader and can pass a superficial fluency review. The failure is semantic: the target version leads the audience to believe a different proposition about an electrical system, an action or an uncertainty.
Repair. Compare assertions rather than just terminology. Compare the disputed statement with its original source field and relevant neighbouring records. Ask whether the reader can still recover the correct actor, time, asset, value, status and authority. If evidence is incomplete, preserve the incompleteness and refer the decision to the responsible specialist rather than creating a confident substitute.
Translation-memory contamination
Signature. A previous project’s valid term is reused in a different system. Why it matters. System-specific statuses are merged incorrectly. The mistake may be invisible to a monolingual reader and can pass a superficial fluency review. The failure is semantic: the target version leads the audience to believe a different proposition about an electrical system, an action or an uncertainty.
Repair. Filter memory by client, domain, version and term status. Compare the disputed statement with its original source field and relevant neighbouring records. Ask whether the reader can still recover the correct actor, time, asset, value, status and authority. If evidence is incomplete, preserve the incompleteness and refer the decision to the responsible specialist rather than creating a confident substitute.
Unreviewed back translation
Signature. A back-translated sentence appears plausible but differs in control semantics. Why it matters. Round-trip fluency is mistaken for proof of source fidelity. The mistake may be invisible to a monolingual reader and can pass a superficial fluency review. The failure is semantic: the target version leads the audience to believe a different proposition about an electrical system, an action or an uncertainty.
Repair. Use source-led technical review and typed data comparisons. Compare the disputed statement with its original source field and relevant neighbouring records. Ask whether the reader can still recover the correct actor, time, asset, value, status and authority. If evidence is incomplete, preserve the incompleteness and refer the decision to the responsible specialist rather than creating a confident substitute.
Missing decision owner
Signature. An unresolved electrical ambiguity is silently settled by a linguist. Why it matters. The translation process takes authority it does not possess. The mistake may be invisible to a monolingual reader and can pass a superficial fluency review. The failure is semantic: the target version leads the audience to believe a different proposition about an electrical system, an action or an uncertainty.
Repair. Record a responsible domain owner and block consequential guesses. Compare the disputed statement with its original source field and relevant neighbouring records. Ask whether the reader can still recover the correct actor, time, asset, value, status and authority. If evidence is incomplete, preserve the incompleteness and refer the decision to the responsible specialist rather than creating a confident substitute.
Worked diagnostics: sixty-four fictional power-system translation cases
The following cases use fictitious network identifiers and deliberately plain-English renderings of possible target-language errors. The goal is to make the distortion visible even to readers who do not speak a particular language pair. A real production translation would be written in the assigned target language and reviewed by a qualified bilingual specialist together with the authorised electrical or utility subject-matter owner. None of these examples authorises live switching or presents a real company’s operational situation.
Read each case in three passes. First identify the asserted source fact, including whether it is observed, provisional, desired or completed. Next compare that source fact with a tempting but misleading target reading; notice precisely what has been added, omitted or altered. Finally evaluate the repaired interpretation and the requested reviewer evidence. The point is not to memorize a stock translation but to learn a portable method for another language, dataset and communication channel.
Numbers, timestamps and references are illustrative. Terms such as open, active, dispatched, unavailable, normal, returned, accepted and cleared often have specialised meanings defined by a particular system. A target string must not assume that all systems use those meanings identically. In an actual assignment, the source’s approved data dictionary and effective system version govern the interpretation. When the source alone does not resolve a status, the case teaches escalation rather than certainty.
Diagnostic 001: Asset identifier versus description
Source situation. The equipment tag TR-A17 is paired with a display label ‘river-side transformer’. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target replaces TR-A17 with a translated nickname. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Keep TR-A17 intact and translate only the human-readable display label. The identifier anchors inspection records, while the display description helps readers orient themselves. The two fields have different translation permissions. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 002: Station name ambiguity
Source situation. A service note lists ‘Station Cedar’ as an official name, while cedar also appears in an ordinary landscape sentence. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. Both appearances are automatically localized as ordinary plant vocabulary. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Retain the approved station name and translate the common noun in the descriptive sentence. A name database, not an unassisted dictionary, determines which token is a reference to an asset. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 003: Transmission versus service supply
Source situation. The source reports a constraint on a transmission corridor without declaring customer interruptions. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target says households have lost electricity. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Report a transmission constraint, and mention customer effects only if supported elsewhere. A network constraint and a service interruption are related possibilities, not equivalent states. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 004: Distribution feeder responsibility
Source situation. The outage record assigns a distribution feeder to operator B and a nearby transmission line to operator A. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target attributes both to operator A. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Preserve each asset’s named owner and operational boundary. An apparently minor actor substitution can redirect responsibility and corrupt later correspondence. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 005: Modeled connection
Source situation. The topology model describes an electrical path between busbar B4 and feeder F6. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target says F6 is presently energised. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Describe the modeled connection without claiming live voltage. The existence of a network edge cannot prove the condition of the connected customer supply. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 006: Snapshot topology
Source situation. An approved network model was exported on 8 October; a live reconfiguration occurred later. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The translated report calls the old model the current arrangement. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. State that it is the topology snapshot from 8 October. The date qualifies whether an apparently accurate diagram still describes reality. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 007: Breaker open indication
Source situation. A remote point reports breaker CB-12 as OPEN at 14:03. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target says a technician safely isolated CB-12. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Write that an open indication was received at 14:03. The indication records state, not the agent, isolation procedure or safety outcome. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 008: Close command queued
Source situation. The event log says CLOSE COMMAND QUEUED and has no positive feedback. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target claims the breaker was closed. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Retain ‘close command queued; execution not confirmed’. The queue state is distinct from command dispatch and from physical feedback. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 009: Disconnector device class
Source situation. The equipment sheet names an off-load disconnector DS-2. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target identifies a circuit breaker with protective tripping capability. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Use the approved disconnector term and its identifier. Different electrical apparatus may share a colloquial label yet perform different functions. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 010: Switch label in customer text
Source situation. A public message says a network switch will be inspected, without specifying its class. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The translation invents a particular protective relay. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Use a general approved description, or query the utility for the intended device. Specificity not present in the source creates a new technical claim. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 011: No observed voltage
Source situation. A technician’s report records ‘0 V at test point TP3’ without declaring the installation safe for work. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target says personnel may touch the equipment. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. State only the recorded zero-voltage observation and its test point. A measurement cannot be converted by a translator into an authorisation or comprehensive safety conclusion. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 012: Administrative de-energised state
Source situation. A work-management platform lists an asset as de-energized but has an incomplete verification checklist. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target labels the circuit proved dead and cleared for access. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Retain the reported status and unresolved verification requirement. System labels can reflect workflow states rather than all evidence necessary for physical safety. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 013: Nominal and actual voltage
Source situation. A data table contains nominal 33 kV and measured 32.6 kV. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target reports an exactly measured 33 kV. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Keep the nominal rating and actual reading as separate fields. Engineering class and observed condition answer different questions. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 014: Phase reference omitted
Source situation. A reading says 230 V line-to-neutral; another value is line-to-line. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target presents both as the same undifferentiated voltage. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Maintain the stated reference for each reading. Identical units do not establish identical measurement configurations. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 015: Power and energy
Source situation. The dispatch note states peak active power of 7 MW. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The translation gives consumption of 7 MWh. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Use active power of 7 MW, and make no energy claim. A rate cannot be converted into accumulated energy without an interval and appropriate integration. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 016: Interval register
Source situation. A metering record shows 16 kWh over a half-hour interval. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target says instantaneous load equals 16 kW. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Report 16 kWh in the specified interval. If a derived average power is needed, it must be authorized and labelled as derived rather than copied. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 017: Reactive support
Source situation. A capacitor bank delivers 2.5 Mvar under one operating condition. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target calls it a generator producing 2.5 MW. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Preserve reactive-power quantity and avoid implying generation of active power. The distinction affects how the equipment’s purpose is understood. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 018: Apparent power rating
Source situation. A transformer nameplate shows 20 MVA while a report lists 14 MW active flow. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target collapses both into one 20 MW rating. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Keep the 20 MVA nameplate and 14 MW active flow distinct. The numerical units belong to different electrical concepts and have different business implications. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 019: Frequency indication
Source situation. A monitor shows 49.94 Hz at one sample time. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target declares the system has fully stabilized. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Report the value and observation time only. A single frequency reading does not establish sustained stability or service availability. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 020: Control-system action
Source situation. A controller recorded an automatic frequency response; no personnel intervention is noted. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target says operators manually corrected frequency. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Preserve automatic response without inventing a human actor. Agency is part of the operational evidence chain. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 021: Available generation
Source situation. A plant record says 80 MW available, 54 MW dispatched. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target claims it produced 80 MW. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Distinguish available capability, dispatch instruction and actual output. Capacity is neither a schedule nor proof of physical production. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 022: Output setpoint
Source situation. A dispatch system issues a 45 MW setpoint but receives no measurement confirmation. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target says the unit generated 45 MW. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. State that a 45 MW setpoint was issued, without asserting achieved output. An instructed target and the measured response are different events. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 023: Demand shed
Source situation. A demand programme reports a 12 MW reduction in load. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target says twelve megawatt-hours of electricity were generated. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Report the decrease in demand as power, not energy production. The wrong subject and wrong dimension create two independent errors. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 024: Renewable curtailment
Source situation. A solar plant receives a limitation on its exported active power. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target tells customers their power is intentionally disconnected. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Describe the generation export constraint. Curtailment at a producer does not necessarily mean interruption to electricity consumers. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 025: Tap change requested
Source situation. Transformer tap position 9 is requested but the actual position still reads 8. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target says tap position 9 is active. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Present requested and observed positions separately. A request can fail or remain pending even when transmitted successfully. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 026: Voltage regulator status
Source situation. The source reports regulator in MANUAL mode with no adjustment carried out. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target says automatic regulation is functioning normally. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Preserve MANUAL mode and absence of adjustment. A target adjective such as normal cannot override a specific source mode. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 027: Protective relay operation
Source situation. A relay event states earth-fault element operated and breaker CB4 tripped. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target asserts a confirmed damaged underground cable. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Report the protection event and do not identify an unverified physical fault. A relay detects conditions; later inspection may be needed to establish cause. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 028: Manual opening record
Source situation. An operator log records a manually initiated opening of CB8. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target calls it an automatic trip. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Retain manually initiated opening and its logged actor. The distinction changes incident chronology and responsibility. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 029: Abnormal current reading
Source situation. A sensor flagged high current above an event threshold. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target reports that an insulation failure was established. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Retain high-current indication and refer cause to investigation. One abnormal parameter may have several possible explanations. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 030: Preliminary cause classification
Source situation. A ticket shows ’cause: under investigation’ with a candidate category. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target prints the candidate as final cause. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Preserve provisional status and the source’s uncertainty. A classification suggestion is not equivalent to an approved incident finding. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 031: Estimated fault section
Source situation. A fault-location tool identifies a likely segment between numbered switches. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target gives one precise pole as the confirmed defect. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Communicate the estimated segment and uncertainty. Model-based location and physical inspection have different evidential strengths. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 032: Affected area boundaries
Source situation. A public draft lists two zones as potentially affected but only one is confirmed. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target says both zones are interrupted. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Preserve the possible-versus-confirmed boundary. A translator must not flatten scope uncertainty into definitive geography. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 033: Maintenance not yet approved
Source situation. A work request for Saturday remains in review. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target announces a scheduled shutdown for Saturday. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Call it a proposed maintenance request awaiting approval. An internal plan does not by itself create a valid customer notice. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 034: Cancelled planned work
Source situation. A planned maintenance notice is superseded by a cancellation. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The translated channel still promotes the previous interruption notice. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Publish the cancellation with the same event identity and withdraw the old notice. Translation correctness also depends on version alignment. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 035: Provisional ETR
Source situation. A customer update says restoration is estimated for 21:00 and subject to assessment. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target promises electricity at exactly 21:00. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Keep the estimate and its stated dependency. Future tense and omission of qualification can accidentally create a guarantee. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 036: Updated ETR
Source situation. A later approved estimate changes from 18:00 to 20:30. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target headline retains the old time but body uses the new one. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Align every display field with the same approved estimate version. An accurate body cannot compensate for a contradictory notification headline. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 037: Partial recovery
Source situation. Three of five feeder sections are restored. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target says the entire service area is restored. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. State the extent and leave two sections outstanding. Scope is part of the fact, not optional fine print. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 038: Residual isolated accounts
Source situation. A main incident is resolved but seven individual connections remain without supply. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target reports zero affected customers. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Preserve the residual group and the difference between network and customer status. Counts often come from distinct systems and cannot be merged casually. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 039: Crew dispatched
Source situation. The task record says crew assigned at 09:10 and en route at 09:20. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target says repair was completed at 09:20. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Report assignment and departure stages only. A workflow chronology can be distorted by attractive but unsupported verbs. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 040: Crew access delayed
Source situation. The source says access is pending an authorised assessment. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The translation describes technicians as already working at the defect. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Maintain access-pending condition without supplying any work instructions. The language should not silently remove a genuine operational dependency. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 041: Telemetry acknowledgement
Source situation. SCADA confirms receipt of a remote-control request. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target describes physical completion of the requested action. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Use ‘request acknowledged’ and require separate execution feedback. A communications handshake is not the same as equipment state change. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 042: Alarm acknowledgement
Source situation. An operator acknowledges an active alarm in a monitoring screen. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target says the fault has been cleared. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Retain ‘alarm acknowledged’ while status remains active. Acknowledgement is a human information workflow, not proof of remediation. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 043: Stale telemetry
Source situation. A voltage point carries its last-good value but a stale-quality flag. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The translated panel shows the value without its stale label. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Keep the value’s age or quality qualification visible. Readers otherwise infer freshness that the source system explicitly rejects. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 044: Unreachable device
Source situation. The gateway reports ‘communication lost’ without a confirmed equipment failure. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target says the device is destroyed. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Report loss of communication, not the physical condition of the device. Absence of telemetry must not become evidence of a particular physical damage. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 045: CIM class name
Source situation. An XML mapping contains a schema class name and a user-facing equipment description. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target translator changes the class name to a localized phrase. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Preserve schema identity and translate the description field only. Machine-recognized class labels serve interoperability, not ordinary reading. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 046: Object reference mismatch
Source situation. A bilingual export keeps each title but accidentally swaps two network object IDs. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The translated document looks plausible but points to the wrong asset. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Reconcile IDs against source records before release. Technical translation includes relational integrity as well as lexical fidelity. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 047: IEC 61850 node code
Source situation. A substation message includes an IEC logical-node code as a machine field. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target version translates the code as a sentence. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Keep the machine code unchanged and add a separate description if authorised. Standards-based identifiers should not be casually naturalised. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 048: Data-object alias
Source situation. A display layer gives an explanatory alias for a device attribute. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The alias is copied into the structured-data field in place of the original key. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Limit translation to the display string. The distinction between user-friendly alias and data field prevents message parsing failures. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 049: Meter cumulative total
Source situation. A smart meter shows a cumulative import register of 2,410 kWh. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target reports this as yesterday’s usage. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Describe the cumulative register, not a daily interval. A time window must be established before a consumption difference can be asserted. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 050: Meter interval clock
Source situation. An interval reading is recorded in UTC while the customer portal displays local clock labels. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The translation suppresses timezone and invents a matching local period. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Retain the recorded UTC interval or perform and label an approved conversion. An hour shifted across a date boundary can change which period the reading appears to cover. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 051: Battery capability
Source situation. A storage installation has an energy capacity of 12 MWh and a power limit of 4 MW. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target says it can supply 12 MW. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Keep energy capacity and power limit separate. Battery duration and instantaneous output cannot be inferred from one rating alone. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 052: Export permission
Source situation. An interconnection agreement allows export only under specified conditions. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target says the device always exports at its nameplate output. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Preserve conditional permission without asserting actual generation. Legal permission, installed capability and measured output are three distinct propositions. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 053: Solar forecast range
Source situation. The forecast says expected generation between 20 and 28 MW for a period. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target gives guaranteed production of 28 MW. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Keep the interval and forecast character. A range describes uncertainty and must not be tightened into its favourable endpoint. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 054: Wind curtailment event
Source situation. A control report says wind generation was curtailed for a stated network condition. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target claims a mechanical turbine failure caused the change. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Report curtailment and documented reason without inventing equipment damage. Dispatch limits and mechanical breakdowns are not synonymous. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 055: Black-start capability
Source situation. The asset register marks a unit ‘black-start capable’. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target reports that the unit has begun energizing a network. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Preserve the capability statement, not an operational completion claim. Potential ability is not the performance of a real event; do not add operational instructions. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 056: Restoration plan approval
Source situation. A restoration plan is being reviewed by an authorised operations body. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target presents it as an approved sequence to be executed. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. State that the plan remains under review. Translation cannot approve a high-risk operational procedure. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 057: Demand-response invitation
Source situation. Eligible customers are invited to opt into a demand programme. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target says all eligible customers have been enrolled. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Keep invitation, eligibility and enrollment apart. Consent and participation are documented customer states, not implied by eligibility. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 058: Reward versus requirement
Source situation. A programme offers a conditional incentive for reducing demand. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target describes a compulsory electricity ration. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Retain the voluntary and conditional nature of the offer. The audience’s decision rights are changed by the wrong modality. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 059: Public outage hypothesis
Source situation. An internal ticket contains a suspected equipment problem and an external update says investigation continues. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The translated public message declares the suspected equipment to be the cause. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Follow the approved public-facing source and its uncertainty. Internal troubleshooting material is not automatically a public fact. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 060: Estimated customer count
Source situation. The operations dashboard reports ‘approximately 800 accounts potentially affected’. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target displays ‘exactly 800 houses without power’. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Keep approximation, affected-unit type and potential status. Accounts, houses, meters and confirmed interruptions represent different measures. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 061: Unexplained gateway outage
Source situation. A control-centre incident reports a gateway unreachable for 14 minutes. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target attributes it to hostile intrusion. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. State the availability incident and defer attribution. A security conclusion requires separate evidence and approval. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 062: Security review pending
Source situation. An investigation is assigned to cybersecurity specialists with no finding issued. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target announces a confirmed cyber incident. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Preserve pending review and distinguish suspicion from conclusion. An assignment of responsibility does not prove the cause under investigation. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 063: Draft versus final report
Source situation. A draft incident review proposes two possible causes. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The target article gives one as the official result. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Keep draft status and alternatives. Document maturity and evidential uncertainty must survive language transfer. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Diagnostic 064: Retrospective correction
Source situation. An investigation later corrects a previously published event time. This is an invented teaching case. A professional reviewer first identifies which part is an asset, which part is an event or measurement, and which part records an uncertainty, purpose or period. That classification prevents an otherwise natural target sentence from acquiring authority that the source never had.
Misleading target-language interpretation. The translation quietly rewrites the original history without a correction note. The rendering could sound smooth and credible to a reader who has not seen the source. Its problem is not necessarily a misspelling or visibly wrong grammar: it changes what the record permits the audience to believe. A fluent sentence that communicates a different condition is still a failed translation.
Repaired interpretation. Publish the revised claim with version history and a correction marker. Readers and auditors need to distinguish the original assertion from the later finding. The repaired translation must remain natural in the actual target language; it need not mirror the English word order. The reviewer should explicitly compare the original’s agent, object, measurement, time, certainty and action status with the final target text, and record any unresolved technical question.
Verification and transfer. Retrieve the source entry and, where necessary, its associated status code, timestamp, quantity definition, note and approval. Ask a bilingual reviewer what they believe happened after reading only the proposed target; then ask the domain owner whether that belief matches the authorised source claim. If not, revise the wording or seek clarification. The useful skill is to preserve the relationship illustrated by this case when another language pair uses very different grammar or terminology.
Transfer workshops: eighteen exercises for a different language, system and audience
The purpose of a transfer workshop is not to repeat the preceding examples. Each exercise changes one variable while keeping others stable, so that the learner becomes less dependent on a memorised phrase. The briefs below are fictional and educational. Use them in translation training, glossary workshops, team review or quality-assurance planning; do not use them as real electrical operating instructions. For every exercise, write a target-language version, a one-sentence fidelity justification and a list of questions that require a domain owner’s answer.
Workshop 01. Rebuild an outage headline
Production brief. An approved source notice says: ‘Interruption affecting approximately 1,200 service accounts; investigation underway; next update expected at 15:00 UTC.’ Create a short public headline and a longer accessible body paragraph without changing the estimate, counting unit, investigation status or timezone. The translator’s deliverable should identify the target audience, document function and approved source version before offering any stylistic choices. As with every case in this guide, all events and figures are fictitious.
How to evaluate the transfer. Test whether the short format still includes the essential uncertainty. Ask a reader whether power is guaranteed by 15:00; the correct answer is no because that is an update time, not an estimated restoration time. Compare whether service accounts have been misdescribed as people, homes or meters. The exercise teaches a valuable distinction between when information will be supplied and when a physical service might resume. A second reviewer should separate the linguistic quality from the technical or evidential fidelity. A stylish answer that strengthens a forecast, changes a dimension or confuses a workflow stage should fail even when its grammar is excellent.
Repeatability question. What principle would still be true if the language direction were reversed, the date changed, the network identifier were different, or the same message were shown in a mobile notification instead of a long report? Record that principle as a reusable test, then compare it with the project termbase and the source-owner’s approval. This turns one worked example into a transferable translation skill rather than a memorized response.
Workshop 02. Defend a numeric table
Production brief. A fictional export lists nominal voltage 33 kV, present indication 32.8 kV, available active power 16.5 MW and interval energy 8.2 MWh. Translate the table headers and add a plain-language note without changing the values or measurements. The translator’s deliverable should identify the target audience, document function and approved source version before offering any stylistic choices. As with every case in this guide, all events and figures are fictitious.
How to evaluate the transfer. Ask reviewers to explain which number is a network class, which is an observation, which is power and which is energy. Never let the explanatory note generate a new result. Test decimal separators, direction, missing units and mobile line wrapping. The exercise forces the translator to treat quantities as typed data, not convenient evidence for a reassuring narrative. A second reviewer should separate the linguistic quality from the technical or evidential fidelity. A stylish answer that strengthens a forecast, changes a dimension or confuses a workflow stage should fail even when its grammar is excellent.
Repeatability question. What principle would still be true if the language direction were reversed, the date changed, the network identifier were different, or the same message were shown in a mobile notification instead of a long report? Record that principle as a reusable test, then compare it with the project termbase and the source-owner’s approval. This turns one worked example into a transferable translation skill rather than a memorized response.
Workshop 03. Preserve command chronology
Production brief. A log shows request submitted at 10:01:05, acknowledgement at 10:01:06 and confirmation unavailable. Translate it for an operations audit and a nontechnical supervisor. The translator’s deliverable should identify the target audience, document function and approved source version before offering any stylistic choices. As with every case in this guide, all events and figures are fictitious.
How to evaluate the transfer. Both audiences must understand that the request was received but successful operation is not established. The audit version may preserve precise timestamps, while the supervisor version can include a clear explanation of why the missing confirmation matters. The key is not to invent a completed action merely because the command workflow advanced one stage. A second reviewer should separate the linguistic quality from the technical or evidential fidelity. A stylish answer that strengthens a forecast, changes a dimension or confuses a workflow stage should fail even when its grammar is excellent.
Repeatability question. What principle would still be true if the language direction were reversed, the date changed, the network identifier were different, or the same message were shown in a mobile notification instead of a long report? Record that principle as a reusable test, then compare it with the project termbase and the source-owner’s approval. This turns one worked example into a transferable translation skill rather than a memorized response.
Workshop 04. Resolve a glossary conflict
Production brief. One customer-facing glossary translates ‘feeder’ as a supply circuit. A legacy engineering manual uses the same target noun for a customer service connection. A new source document includes both concepts. The translator’s deliverable should identify the target audience, document function and approved source version before offering any stylistic choices. As with every case in this guide, all events and figures are fictitious.
How to evaluate the transfer. Build two concept records with definitions, identifiers, permitted audience variants and examples. Show how you would flag translation-memory matches from the older manual and ask the engineering owner for an approved distinction. The deliverable is a terminological decision log, not an arbitrary selection of the shorter synonym. A second reviewer should separate the linguistic quality from the technical or evidential fidelity. A stylish answer that strengthens a forecast, changes a dimension or confuses a workflow stage should fail even when its grammar is excellent.
Repeatability question. What principle would still be true if the language direction were reversed, the date changed, the network identifier were different, or the same message were shown in a mobile notification instead of a long report? Record that principle as a reusable test, then compare it with the project termbase and the source-owner’s approval. This turns one worked example into a transferable translation skill rather than a memorized response.
Workshop 05. Separate forecast and observation
Production brief. A solar operations report contains yesterday’s actual output and tomorrow’s expected generation range. Both tables use MW. Translate a narrative that compares them without implying either forecast is a measured value. The translator’s deliverable should identify the target audience, document function and approved source version before offering any stylistic choices. As with every case in this guide, all events and figures are fictitious.
How to evaluate the transfer. The source’s grammatical tense is insufficient by itself: the table headings and dates establish the evidential type. Have a reviewer identify every forecast, actual, issue time and period. A good translation supports comparison while preserving uncertainty and refrains from implying that a planned or forecast value was achieved. A second reviewer should separate the linguistic quality from the technical or evidential fidelity. A stylish answer that strengthens a forecast, changes a dimension or confuses a workflow stage should fail even when its grammar is excellent.
Repeatability question. What principle would still be true if the language direction were reversed, the date changed, the network identifier were different, or the same message were shown in a mobile notification instead of a long report? Record that principle as a reusable test, then compare it with the project termbase and the source-owner’s approval. This turns one worked example into a transferable translation skill rather than a memorized response.
Workshop 06. Create a partial restoration notice
Production brief. The fictional event begins with 500 affected accounts. Supply returns to 430; the other 70 remain interrupted while assessment continues. Write an update in a target language with explicit accessibility requirements. The translator’s deliverable should identify the target audience, document function and approved source version before offering any stylistic choices. As with every case in this guide, all events and figures are fictitious.
How to evaluate the transfer. The notice needs an accurate affected balance, a plain-language explanation and no premature ‘all clear’. Review whether the 430 refers to accounts or customers and whether any estimate is present. Test how the notice behaves as an SMS, a webpage banner and a screen-reader announcement. The exercise teaches scope integrity under interface constraints. A second reviewer should separate the linguistic quality from the technical or evidential fidelity. A stylish answer that strengthens a forecast, changes a dimension or confuses a workflow stage should fail even when its grammar is excellent.
Repeatability question. What principle would still be true if the language direction were reversed, the date changed, the network identifier were different, or the same message were shown in a mobile notification instead of a long report? Record that principle as a reusable test, then compare it with the project termbase and the source-owner’s approval. This turns one worked example into a transferable translation skill rather than a memorized response.
Workshop 07. Test a negative statement
Production brief. The source says ‘No evidence of a malicious incident has been established at this time.’ Translate it without turning it into either proof of an attack or proof that no attack could have occurred. The translator’s deliverable should identify the target audience, document function and approved source version before offering any stylistic choices. As with every case in this guide, all events and figures are fictitious.
How to evaluate the transfer. Check the scope of the negative, the temporal phrase and the evidential verb. A target may need a fuller construction to preserve the difference between absence of established evidence and established absence. The review should be done by the authorized incident communication owner and the bilingual editor, not by a speculative inference. A second reviewer should separate the linguistic quality from the technical or evidential fidelity. A stylish answer that strengthens a forecast, changes a dimension or confuses a workflow stage should fail even when its grammar is excellent.
Repeatability question. What principle would still be true if the language direction were reversed, the date changed, the network identifier were different, or the same message were shown in a mobile notification instead of a long report? Record that principle as a reusable test, then compare it with the project termbase and the source-owner’s approval. This turns one worked example into a transferable translation skill rather than a memorized response.
Workshop 08. Distinguish network restoration from work completion
Production brief. A field ticket says a damaged component has been replaced, but restoration tests are not complete. Compose an internal handover summary and a customer notice. The translator’s deliverable should identify the target audience, document function and approved source version before offering any stylistic choices. As with every case in this guide, all events and figures are fictitious.
How to evaluate the transfer. The internal document may mention the repair milestone; the external document must not promise supply until validated. Design one checklist for each audience and explain why the documents share event identity but have different detail levels. The translation should be truthful in both places, with no hidden leap between repairing equipment and restoring service. A second reviewer should separate the linguistic quality from the technical or evidential fidelity. A stylish answer that strengthens a forecast, changes a dimension or confuses a workflow stage should fail even when its grammar is excellent.
Repeatability question. What principle would still be true if the language direction were reversed, the date changed, the network identifier were different, or the same message were shown in a mobile notification instead of a long report? Record that principle as a reusable test, then compare it with the project termbase and the source-owner’s approval. This turns one worked example into a transferable translation skill rather than a memorized response.
Workshop 09. Protect names across scripts
Production brief. A dataset includes station code SUB-W7, an approved public place name, a staff surname and the ordinary word ‘station’ in several sentences. Translate the explanatory prose for a writing system unlike the source. The translator’s deliverable should identify the target audience, document function and approved source version before offering any stylistic choices. As with every case in this guide, all events and figures are fictitious.
How to evaluate the transfer. Decide which elements are copied as codes, which have an approved official target-language form, which require transliteration and which are common nouns. Present your naming policy as a small table with supporting evidence. A back-translation alone is not enough because visually different scripts can refer to the same entity or to entirely different entities. A second reviewer should separate the linguistic quality from the technical or evidential fidelity. A stylish answer that strengthens a forecast, changes a dimension or confuses a workflow stage should fail even when its grammar is excellent.
Repeatability question. What principle would still be true if the language direction were reversed, the date changed, the network identifier were different, or the same message were shown in a mobile notification instead of a long report? Record that principle as a reusable test, then compare it with the project termbase and the source-owner’s approval. This turns one worked example into a transferable translation skill rather than a memorized response.
Workshop 10. Review interval billing language
Production brief. A meter report says 1.8 kWh import over a 30-minute interval and 2.2 kWh export over the following interval, using a documented sign convention. Explain this in natural target-language prose. The translator’s deliverable should identify the target audience, document function and approved source version before offering any stylistic choices. As with every case in this guide, all events and figures are fictitious.
How to evaluate the transfer. Keep the two intervals separate and retain the original report’s definitions of import and export. Avoid averaging or netting readings without a specifically authorized calculation. Ask whether a reader can reconstruct the direction, amount and time window of each recorded exchange. This exercise highlights the danger of using one casual word for several energy quantities. A second reviewer should separate the linguistic quality from the technical or evidential fidelity. A stylish answer that strengthens a forecast, changes a dimension or confuses a workflow stage should fail even when its grammar is excellent.
Repeatability question. What principle would still be true if the language direction were reversed, the date changed, the network identifier were different, or the same message were shown in a mobile notification instead of a long report? Record that principle as a reusable test, then compare it with the project termbase and the source-owner’s approval. This turns one worked example into a transferable translation skill rather than a memorized response.
Workshop 11. Diagnose a machine-translation invention
Production brief. A source fault note reads ‘Alarm acknowledged; cause pending investigation.’ A translation system returns ‘The damaged transformer has been repaired and the alarm resolved.’ Write an error report that can be used for training and vendor feedback. The translator’s deliverable should identify the target audience, document function and approved source version before offering any stylistic choices. As with every case in this guide, all events and figures are fictitious.
How to evaluate the transfer. Classify the invented asset, invented diagnosis, invented action and invented completion status separately. Explain which source tokens or system fields would disprove each addition. The reviewer should propose a faithful target equivalent and recommend a targeted test set. Correcting the one sentence is useful; preventing a recurring hallucination is more valuable. A second reviewer should separate the linguistic quality from the technical or evidential fidelity. A stylish answer that strengthens a forecast, changes a dimension or confuses a workflow stage should fail even when its grammar is excellent.
Repeatability question. What principle would still be true if the language direction were reversed, the date changed, the network identifier were different, or the same message were shown in a mobile notification instead of a long report? Record that principle as a reusable test, then compare it with the project termbase and the source-owner’s approval. This turns one worked example into a transferable translation skill rather than a memorized response.
Workshop 12. Build a source-version comparison
Production brief. A notice at version 2 estimated restoration at 16:00; version 3 gives 18:30, changes the affected count and explicitly supersedes version 2. Plan a synchronized bilingual publication. The translator’s deliverable should identify the target audience, document function and approved source version before offering any stylistic choices. As with every case in this guide, all events and figures are fictitious.
How to evaluate the transfer. Prepare a change record with source ID, both versions, affected languages and withdrawal requirements. Test that old notification banners do not remain visually dominant. A reader reaching a cached translation must see whether it is current or superseded. This exercise teaches translation as change management in a fast-moving information system. A second reviewer should separate the linguistic quality from the technical or evidential fidelity. A stylish answer that strengthens a forecast, changes a dimension or confuses a workflow stage should fail even when its grammar is excellent.
Repeatability question. What principle would still be true if the language direction were reversed, the date changed, the network identifier were different, or the same message were shown in a mobile notification instead of a long report? Record that principle as a reusable test, then compare it with the project termbase and the source-owner’s approval. This turns one worked example into a transferable translation skill rather than a memorized response.
Workshop 13. Move from model to interface
Production brief. A machine exchange uses a structured object ID, a status enum and a localized description field. Translate the interface without modifying the underlying structure. The translator’s deliverable should identify the target audience, document function and approved source version before offering any stylistic choices. As with every case in this guide, all events and figures are fictitious.
How to evaluate the transfer. State exactly which fields are editable by translators and which are protected by schema rules. Show a before-and-after fixture where the numeric and identity data are byte-stable but the display language changes. Demonstrate how an automated parse test and a human language review evaluate different aspects of the same deliverable. A second reviewer should separate the linguistic quality from the technical or evidential fidelity. A stylish answer that strengthens a forecast, changes a dimension or confuses a workflow stage should fail even when its grammar is excellent.
Repeatability question. What principle would still be true if the language direction were reversed, the date changed, the network identifier were different, or the same message were shown in a mobile notification instead of a long report? Record that principle as a reusable test, then compare it with the project termbase and the source-owner’s approval. This turns one worked example into a transferable translation skill rather than a memorized response.
Workshop 14. Translate an uncertainty chain
Production brief. A customer complaint suggests an interruption; a remote meter is unreachable; a field visit is requested; no root cause has been established. Write a concise incident chronology. The translator’s deliverable should identify the target audience, document function and approved source version before offering any stylistic choices. As with every case in this guide, all events and figures are fictitious.
How to evaluate the transfer. Label each event by its source and evidence quality. The complaint is not a confirmed network-wide outage, the missing meter communication is not necessarily loss of supply, and the requested visit is not proof of arrival. Ask the target reader what is known versus planned. If the translation collapses the chain into one confirmed failure, it has lost the incident’s actual informational structure. A second reviewer should separate the linguistic quality from the technical or evidential fidelity. A stylish answer that strengthens a forecast, changes a dimension or confuses a workflow stage should fail even when its grammar is excellent.
Repeatability question. What principle would still be true if the language direction were reversed, the date changed, the network identifier were different, or the same message were shown in a mobile notification instead of a long report? Record that principle as a reusable test, then compare it with the project termbase and the source-owner’s approval. This turns one worked example into a transferable translation skill rather than a memorized response.
Workshop 15. Write a multilingual style-sheet entry
Production brief. A team repeatedly mishandles the words ‘scheduled’, ‘estimated’, ‘confirmed’ and ‘restored’. Build four entries with definitions, preferred target forms, prohibited synonyms and two example sentences each. The translator’s deliverable should identify the target audience, document function and approved source version before offering any stylistic choices. As with every case in this guide, all events and figures are fictitious.
How to evaluate the transfer. The style sheet should include audience and source constraints, not just word pairs. For ‘restored’, require an explicit subject: what service or equipment was restored, for whom and when? For ‘confirmed’, specify who confirmed what. The team should be able to use the resource to make a new decision rather than blindly paste the same word into every sentence. A second reviewer should separate the linguistic quality from the technical or evidential fidelity. A stylish answer that strengthens a forecast, changes a dimension or confuses a workflow stage should fail even when its grammar is excellent.
Repeatability question. What principle would still be true if the language direction were reversed, the date changed, the network identifier were different, or the same message were shown in a mobile notification instead of a long report? Record that principle as a reusable test, then compare it with the project termbase and the source-owner’s approval. This turns one worked example into a transferable translation skill rather than a memorized response.
Workshop 16. Reconcile operational and public language
Production brief. An engineering note reads ‘FDR-C2 open due to protective trip; fault cause not established’. A customer statement says ‘Service interrupted in part of the supply area; teams assessing’. Translate both for the same event. The translator’s deliverable should identify the target audience, document function and approved source version before offering any stylistic choices. As with every case in this guide, all events and figures are fictitious.
How to evaluate the transfer. Show a stable relationship between the specialist detail and the public-facing statement without treating one as the complete substitute for the other. Check that both share incident identity, and that public prose avoids accidentally claiming that the technical cause is known. The exercise teaches controlled simplification rather than technical jargon duplication. A second reviewer should separate the linguistic quality from the technical or evidential fidelity. A stylish answer that strengthens a forecast, changes a dimension or confuses a workflow stage should fail even when its grammar is excellent.
Repeatability question. What principle would still be true if the language direction were reversed, the date changed, the network identifier were different, or the same message were shown in a mobile notification instead of a long report? Record that principle as a reusable test, then compare it with the project termbase and the source-owner’s approval. This turns one worked example into a transferable translation skill rather than a memorized response.
Workshop 17. Build a quality test for numbers
Production brief. A translation deliverable contains fifty event rows with values, decimal separators, signed MW flows and ISO timestamps. Design a test plan that detects changed numbers without rejecting legitimate punctuation differences. The translator’s deliverable should identify the target audience, document function and approved source version before offering any stylistic choices. As with every case in this guide, all events and figures are fictitious.
How to evaluate the transfer. Separate raw numeric values from formatted display strings; define which conversions are authorized; audit signs, units, upper and lower bounds, and timezone rendering. Explain why a plain text diff can identify a change but cannot always decide whether a dimensional conversion is justified. The deliverable is a typed comparator specification and a bilingual review policy. A second reviewer should separate the linguistic quality from the technical or evidential fidelity. A stylish answer that strengthens a forecast, changes a dimension or confuses a workflow stage should fail even when its grammar is excellent.
Repeatability question. What principle would still be true if the language direction were reversed, the date changed, the network identifier were different, or the same message were shown in a mobile notification instead of a long report? Record that principle as a reusable test, then compare it with the project termbase and the source-owner’s approval. This turns one worked example into a transferable translation skill rather than a memorized response.
Workshop 18. Prepare an editorial correction
Production brief. A published translated notice incorrectly said ‘fully restored’ when the source said ‘restoration continuing’. The source remains correct. Write a correction notice and an incident learning record. The translator’s deliverable should identify the target audience, document function and approved source version before offering any stylistic choices. As with every case in this guide, all events and figures are fictitious.
How to evaluate the transfer. Identify the false inference, the affected audience, the superseded target version and the repaired wording. Do not imply that an engineering error occurred when the error was linguistic. Keep the amendment visible enough for affected readers to understand the change, and show how the termbase, review checklist and publishing workflow will prevent the same mistake from recurring. A second reviewer should separate the linguistic quality from the technical or evidential fidelity. A stylish answer that strengthens a forecast, changes a dimension or confuses a workflow stage should fail even when its grammar is excellent.
Repeatability question. What principle would still be true if the language direction were reversed, the date changed, the network identifier were different, or the same message were shown in a mobile notification instead of a long report? Record that principle as a reusable test, then compare it with the project termbase and the source-owner’s approval. This turns one worked example into a transferable translation skill rather than a memorized response.
Controlled glossary: seventy-two electrical and translation terms
This glossary offers working explanations rather than formal quotations from proprietary standards. When a contract or system claims adherence to a particular IEC standard or local rule, the exact official definition and adopted edition should be consulted. The central translation lesson is to keep concept identity, measurement type, document function and evidential strength separate.
Active power
The rate at which electrical energy is transferred or converted in the active-power sense; commonly displayed in W, kW or MW. Translation note: Do not replace its unit with an energy unit such as MWh. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Apparent power
A power-system quantity commonly expressed in VA, kVA or MVA. Translation note: Do not silently treat the nameplate MVA rating as measured active MW. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Reactive power
A power-system quantity often represented in var, kvar or Mvar. Translation note: Distinguish from active power and respect the project’s sign convention. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Electrical energy
An accumulated quantity over a period, typically expressed in Wh, kWh or MWh. Translation note: Require an interval or register context when interpreting consumption. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Voltage
Electric potential difference expressed in volts or a multiple such as kV. Translation note: Preserve measured versus nominal value and the reference point. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Current
Electric current, usually measured in amperes. Translation note: Do not swap with voltage or infer missing measurement location. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Frequency
Frequency of alternating electrical quantities, commonly hertz. Translation note: A single observation is not necessarily proof of sustained stability. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Power factor
A performance-related relationship associated with active and apparent power. Translation note: Do not represent a dimensionless factor as a MW quantity. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Nominal voltage
A designated voltage class or reference used to identify a system. Translation note: Do not present as an exactly measured instantaneous value. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Per unit
A normalized representation relative to a defined base. Translation note: Preserve base quantity information and avoid treating the value as an absolute unit. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Feeder
A network circuit or branch used in supply and distribution contexts. Translation note: Check the system definition and distinguish from individual customer service. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Busbar
A conductor arrangement connecting multiple circuits or equipment. Translation note: Preserve device identity and topology, not just the idea of a ‘line’. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Circuit breaker
A switching device with specific protective and switching functions. Translation note: Do not equate automatically with every kind of switch. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Disconnector
An isolating-type device with its own applicable function and restrictions. Translation note: A translator does not grant operating permission by selecting this word. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Fuse
An overcurrent protective device with particular physical characteristics. Translation note: Do not rewrite as an intelligent protective relay or breaker. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Recloser
An automatic switching/protection device often used on distribution networks. Translation note: Keep the equipment class distinct from a generic reconnection event. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Protective relay
A device or function monitoring conditions and initiating protection actions. Translation note: A relay operation may not establish a physical root cause. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Trip
A protective operation or event in a relevant equipment context. Translation note: Distinguish automatic trip from manual open command. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Open indication
A reported open state of a device or contact. Translation note: Does not by itself prove comprehensive safe isolation. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Close request
A requested transition toward a closed equipment state. Translation note: Not equal to a confirmed final state. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Feedback
An observed or returned indication about an operation or process. Translation note: Clarify which device or system generated the feedback. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
SCADA
Supervisory Control and Data Acquisition, used for monitoring and certain control functions. Translation note: Separate observations from instructions and check quality and time. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Telemetry
Remotely acquired measurements or states sent from equipment or systems. Translation note: Do not turn missing telemetry into confirmed equipment damage. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Alarm
A notification that a system has detected a configured condition. Translation note: Acknowledging an alarm need not clear its underlying condition. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Event
A recorded occurrence associated with one or more objects and times. Translation note: Preserve event ID, timestamps and source attribution. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
State estimation
A model-based estimate of system state derived from available data. Translation note: Estimated values are not identical to direct measurements. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Topology
The represented connectivity among network elements. Translation note: A modeled connection is not always confirmed energized service. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Load
Demand or consumption-related electrical quantity, depending on context. Translation note: Require dimensions and time basis rather than guessing. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Demand
An electrical loading or required power quantity over a defined frame. Translation note: Distinguish from cumulative energy and customer count. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Peak demand
A maximum demand within a defined period or measurement basis. Translation note: Preserve the period and quantity units. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Generation
Electric-power production or equipment capability depending on the statement. Translation note: Separate availability, schedule and measured output. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Dispatch
Assignment or instruction affecting the operation of energy resources. Translation note: An instruction does not prove an output was achieved. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Setpoint
A target value requested of a control system or device. Translation note: Preserve its status as a target rather than an actual reading. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Curtailment
A restriction or reduction of an otherwise available energy activity. Translation note: Identify whether production, export or demand is affected. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Demand response
A programme or action involving changes to electricity demand. Translation note: Keep consent and incentives distinct from compulsory interruption. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Outage
Loss or interruption of supply or equipment availability in a defined context. Translation note: Preserve affected scope and distinguish types of outage. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Planned interruption
A supply interruption associated with authorised scheduling. Translation note: An unapproved maintenance request is not a planned customer outage. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Unplanned interruption
An interruption not scheduled as a routine planned event. Translation note: Do not infer a confirmed failure cause from this classification. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Restoration
Return of a relevant equipment or service condition after interruption. Translation note: Name what was restored and whether the scope is partial or complete. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Estimated restoration time
A forecast for when supply or another state may return. Translation note: Keep its estimated, revisable character. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Partial restoration
A return of service to some but not all affected connections. Translation note: Preserve outstanding customers or network segments. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Affected account
A customer-account unit included in an interruption estimate. Translation note: Not necessarily one household, meter or individual. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Customer ticket
A record of a reported customer concern or service request. Translation note: A reported problem is not automatically a verified network fault. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Fault
An abnormal electrical condition defined by a technical context. Translation note: Do not assume a root cause from a general fault label. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Suspected fault
A possible fault under assessment. Translation note: Preserve provisional evidential status. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Fault location
A place or network segment associated with an abnormal condition. Translation note: Distinguish estimate from field-verified position. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Protective operation
An action or event of protection equipment. Translation note: Keep mechanism separate from later causal diagnosis. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Work order
A document or record organizing assigned work. Translation note: May not establish completion or safety approval. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Dispatch status
Information about assignment or movement of personnel or resources. Translation note: En route is different from onsite or work completed. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Work permit
A formal workplace-authorisation concept subject to applicable rules. Translation note: Translation must not create or expand permission. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Safety clearance
A formally established authorization or condition under a defined scheme. Translation note: Never infer it solely from a single voltage indication. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Maintenance window
A period reserved or requested for particular work. Translation note: Preserve whether the window is proposed or approved. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Isolation
A concept concerning separation or removal of an energy pathway under specified controls. Translation note: Avoid substituting it for a casual open indication. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Asset register
Structured information identifying physical or logical equipment. Translation note: Keep stable keys and version history. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Common Information Model
The IEC-associated semantic model used in electric utility data exchange. Translation note: Distinguish human-friendly labels from schema identifiers. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
CIM class
A class or object type in a Common Information Model representation. Translation note: Do not arbitrarily translate it inside an interoperable payload. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
IEC 61970
IEC standard series addressing energy-management application interfaces. Translation note: Refer to a verified edition and relevant part, not an unspecified ‘latest’ standard. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
IEC 61968
IEC standard series addressing distribution-management system interfaces. Translation note: Do not assume every field in a proprietary implementation is normative. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
IEC 61850
IEC standard family for communication networks and systems in power utility automation. Translation note: Preserve machine-level identifiers and verify installed profiles. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Logical node
A functional information-model element in IEC 61850 contexts. Translation note: A logical-node code is not a prose sentence. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Data object
A structured information item associated with a model or device. Translation note: Separate code representation from localised display description. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
AMI
Advanced metering infrastructure, including smart-meter communication functions. Translation note: Do not confuse meter reachability with supply availability. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Interval reading
A measurement or quantity for a defined period. Translation note: Preserve start, end, timezone and measurement type. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Cumulative register
A meter value that accumulates across a continuing period. Translation note: Do not recast as daily usage without a derived difference. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Distributed energy resource
A generating, storage or controllable load resource connected within a network. Translation note: Distinguish capability, dispatch, actual flow and permission. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Battery energy capacity
A storage-energy rating commonly given in kWh or MWh. Translation note: Not interchangeable with maximum MW power. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Export limit
A restriction on power or energy exported under a defined framework. Translation note: Could be contractual, technical, temporary or operational; identify which. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Data quality flag
Metadata indicating validity, staleness or other properties of a value. Translation note: Must remain visible when it changes interpretation. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Source version
The specific authoritative state of a document or dataset used for translation. Translation note: Keep language versions aligned with the same source. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Translation memory
A repository of prior paired segments used for suggestions or reuse. Translation note: Past fluency does not prove present technical compatibility. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Termbase
A structured store of approved concepts, terms, definitions and context. Translation note: Use concept-level control and source-owner authority. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Bilingual review
Comparison of source and target by an appropriately qualified language reviewer. Translation note: Assess fidelity, not only elegance. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Technical sign-off
Approval of technical fidelity by an authorised subject-matter owner. Translation note: Different from permission for physical operations. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Supersession
Replacement of an older version by a later authoritative statement. Translation note: Prior versions should not masquerade as the current notice. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Audit trail
Evidence of versions, decisions, reviewers and relevant changes. Translation note: A traceable claim chain makes corrections defensible. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Public advisory
An authorised message intended for external understanding or action. Translation note: Its source scope and cautions must survive simplification. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Quality assurance
Systematic checks designed to prevent or detect translation failures. Translation note: Separate linguistic, semantic, numerical and technical tests. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Back translation
A rendering of target text back toward the source language. Translation note: Useful as a diagnostic, not definitive evidence of correctness. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Translation ambiguity
Unresolved alternatives in the meaning or role of source content. Translation note: Escalate instead of invisibly guessing a consequential fact. The recommended bilingual equivalent depends on the authorised document, audience and system context, and should be checked against the maintained termbase before publication.
Frequently asked questions about electric power grid translation
What is electricity grid operations translation?
It is the cross-language transfer of information about electric-power systems, operational states, measurements, decisions and service communication. The translator must preserve what is asserted and by whom, as well as the numbers, technical identity and time boundaries. The task goes beyond making electrical jargon readable: it also tests whether the target message implies any unsupported action, promise or cause.
Is translating an outage notice the same as translating a technical manual?
No. A public outage notice prioritizes clear service scope, update time, safety warnings, uncertainty and accessible wording. A technical manual needs exact terminology, procedures as authorised, document configuration and repeatable references. Both demand fidelity, but they have different audiences, approved sources and consequences if a qualifier is omitted.
How do I translate ‘power restored’ accurately?
First ask what has been restored and how completion was established. The phrase might refer to an energized network section, restored equipment functionality or actual customer service. In target languages that require more explicit subjects or completed aspect, that context is essential. Avoid declaring all customers restored unless the source supports that breadth.
Why is an estimated restoration time not a deadline?
An estimate is a forecast formed from incomplete and changeable operational information. A deadline or guarantee asserts a commitment that may not exist. Translate the value, its basis and its update status as the source describes them. A reader should be able to tell whether the time is a service forecast, next information update or confirmed schedule.
Can a machine translator handle IEC terminology?
It can propose equivalents, but it cannot be assumed to know the exact standard edition, client-specific model or live data context. Machine output requires source-led review, protected field handling and technical sign-off when consequential. Schema codes should not be exposed to unconstrained language rewriting. Fluency is one quality dimension, not evidence of operational fidelity.
What is the difference between MW and MWh in translation?
MW is a unit of power; MWh is a unit of energy. Power describes a rate, while energy relates to accumulation over time. They should remain separate even if the target language uses a broad everyday word for electricity. Numbers and units must be validated together and the measurement period retained where applicable.
How should switch and breaker terms be translated?
Use the source’s approved device classification and technical definition. Casual synonyms are not reliable across equipment types. A general public notice may use an authorised broader phrase, while an engineering record must retain the exact class. When the source is ambiguous, consult the equipment register or domain owner rather than guessing from appearance.
Can a translated notice mention the probable cause?
Only if its approved source and disclosure policy support that probability claim. A report that investigation is underway must not become a definitive root-cause statement. Even a familiar mechanism is not evidence of what occurred in this particular event. Keep the confidence and the responsible source attached to any causal language.
What does a SCADA alarm acknowledgement mean?
It typically describes a human or system acknowledgement of an alarm notification in an operational workflow. It does not automatically prove that the underlying condition has disappeared. The exact semantics depend on the system and alarm implementation. The target should reflect the recorded action instead of silently using ‘resolved’.
Should equipment identifiers be translated?
Usually the stable machine or asset identifier should be preserved unchanged unless an authorised mapping says otherwise. Human-facing labels may have approved variants, and a translation can explain a code in prose. Maintain a cross-language entity register so that records can be reliably linked, inspected and corrected.
How do I handle local time and UTC?
Identify which clock the source uses, then translate the date and time without losing the timezone. If an authorised transformation is required, label the resulting time basis and validate date changes, daylight-saving rules and effective intervals. Never assume a target audience will infer UTC correctly from a bare number.
What if the source does not explain a term?
Find the data dictionary, referenced standard, asset record, event context or source owner. Document remaining ambiguity. A high-stakes translation should not invent a convenient precise interpretation; it can preserve uncertainty or use an explicit approved explanation. The decision trail is part of the deliverable.
Are IEC standards themselves bilingual dictionaries?
Some IEC resources provide formal terminology and multilingual equivalents, notably the International Electrotechnical Vocabulary. Other standards provide information models, interfaces or communication structures rather than general bilingual word lists. Verify the actual document, edition and licensing constraints. Do not pretend a standard has approved a translator’s particular wording unless that is documented.
Does the Common Information Model translate human language automatically?
No. CIM supplies shared semantics and object relationships for utility-data interoperability. A human-readable description still requires interpretation and translation; machine identifiers and field names may remain fixed. CIM helps establish concept identity, which can improve translation fidelity, but does not replace review of audience, style and source assertions.
How can a team evaluate machine translation hallucinations?
Compare every target assertion to the source, not just every word. Flag invented assets, causes, crews, times, outcomes, obligations or approvals. Pair bilingual review with typed numeric checks and controlled terminology. Preserve representative failure cases as regression tests for future tool versions and projects.
What is the safest way to shorten a customer update?
Identify the irreducible facts first: who or what is affected, status, uncertainty, next update and any authoritative safety message. Remove background detail before removing meaning-bearing qualifiers. Test the real display, because a channel may cut off the last words. A short notice that drops ‘estimated’ can be materially less accurate than a long one.
Can translated power outage notices use local examples?
A world-facing article can teach the principle with fictional examples. For an actual published notice, local terminology, law and official emergency guidance must come from the responsible operator or authority. Localization adapts legitimate conventions; it does not create unsupported geographical details or operational directions.
How do I preserve partial restoration?
State the affected and restored groups and any remaining exceptions using the same counting unit as the source. If the source counts accounts, do not silently translate that as people or buildings. A message about a major feeder being restored may still require an explicit caveat about connections that remain without supply.
What should be done when source and target disagree?
Stop release of the disputed consequential statement, record the discrepancy and ask the correct owner to settle the source meaning. The bilingual reviewer can describe the language ambiguity; the technical owner decides the engineering fact. Once resolved, update the target, termbase, tests and affected publications with a traceable approval.
What is the relationship between translation and grid safety?
Translation can help preserve an authorised warning or state description, but it is not a safety clearance, electrical qualification or operating permission. Safety-critical records require competent, accountable review. This educational guide deliberately does not teach live switching, energization, isolation or restoration procedures.
Can automatic back translation prove a notice is correct?
No. The reverse translation may sound close while preserving a wrong assumption in both directions. It can reveal ambiguity, but the stronger test is source-led claim comparison, domain validation, quantity parity and comprehension by the intended target audience. Back translation is a supporting diagnostic, not a certificate.
How do translation memories become risky?
A previously approved segment may come from a different utility, technical standard edition, equipment class or document purpose. An identical source word may not carry identical system meaning. Filter memories by domain and version; mark forbidden variants; and require contextual checks before reuse in consequential material.
How often should a live outage notice be translated again?
Whenever its authoritative source changes in a meaning-bearing way and the target version would otherwise become stale. The release workflow should track source revision IDs, translations and withdrawal notices. There is no universal interval that substitutes for the event’s actual update and supersession rules.
What does a successful technical translation look like?
A reader in the target language identifies the same source entity, state, amount, time, confidence and responsibility, and can distinguish what is known from what is planned or estimated. The structured data still links to its original record, and technical reviewers can reconstruct why each high-impact wording decision was approved.
Is this article a substitute for IEC publications or utility procedures?
No. It teaches the translation mechanism and diagnostic habits. IEC documents, specific implementation profiles, employer instructions and competent authorities retain ownership of normative technical requirements and operating rules. When precise compliance or safety interpretation is needed, consult the actual applicable source and an authorized expert.
How can an educator teach this material?
Give learners short fictional records with an incorrect and corrected interpretation, then ask them to explain the difference in certainty, object identity, units, time or agency. Progress from isolated cases to versioned multilingual notices and data-bound translation. Measure their ability to preserve relationships rather than memorizing a list of English electrical terms.
How should an organization document a translation correction?
Record what was wrong, which target version and audience were affected, what source actually supported, who approved the repair, and how superseded copies will be treated. If the error was linguistic rather than engineering, say so accurately. Add a targeted regression test and update the termbase to prevent recurrence.
What belongs in the protected Vocabulary Learning Hub instead?
The acquisition of general word meanings, retrieval, definitions, practice and broad vocabulary development belongs in the Vocabulary Learning Hub. This child focuses on cross-language equivalence when electrical terms are attached to operational evidence and structured claims. It links outward when a reader needs word learning rather than a grid-specific translation decision.
What belongs in How English Works instead?
General English grammar, clause structure, sentence clarity, register and writing mechanisms are owned by How English Works. This child applies those capabilities when an English structure must be carried into another language or reconstructed from another language without changing electrical meaning. The distinction prevents duplicate articles competing for the same basic learning intent.
Research notes and high-authority reference routes
The references below were reviewed on 30 October 2026 for terminology, information-model scope and communication principles. They are not reproduced as standards text; consult the publisher for the full normative language and edition status. Technical standard editions matter because vendors and utilities may implement only particular profiles. Always distinguish an authoritative publisher description from a claim that an individual document or dataset has passed a compliance assessment.
IEC 61970-301:2020+A1:2022, Common Information Model base. Describes the IEC CIM semantic base and its role in interoperable power-system resource information.
IEC 61968-3:2021, Interface for network operations. Explains distribution operations information exchange including topology, control equipment status, complaint handling and fault restoration.
IEC 61968-1:2020, Interface architecture and recommendations. Provides the wider architecture for distribution-management system interfaces.
IEC 61850-7-4 consolidated 2010+A1:2020. Addresses logical node and data-object class information models in utility automation.
IEC 61850 communication concepts. Explains the multiple dimensions of utility interoperability and information exchange.
IEC Electropedia, International Electrotechnical Vocabulary. An official concept-level terminology resource; check the individual entry and subject field.
United States Department of Energy, Staying Informed. Provides an official example of public outage-information guidance and the importance of messages from responsible authorities.
ISO 17100:2015, Translation services. Reference for professional translation processes, distinct from electrical operating standards.
ISO 5060:2024, Translation output evaluation. A reference point for assessing translation output rather than treating surface fluency as sufficient.
Next routes: the translation root and protected subject owners
Return to the parent system: Master Art of Translation | The Complete System for Moving Meaning Between Languages. That root establishes the general architecture of fidelity, equivalence, context, terminology, revision and human judgment. This 560 child is intentionally narrower: it owns the movement of operational electricity-grid meaning between languages, and it does not replace the engineering, safety or business owners of those facts.
For building word knowledge, retrieval and everyday meanings, follow the protected Vocabulary Learning Hub. Use that route when the problem is not knowing a word or being unable to retrieve it. Return to this article when the same word has several possible equivalents and the correct one depends on a technical object, status or source relationship.
For understanding English syntax, clause structure, tone, register and readable expression as English-language capabilities, use the protected How English Works route. This specialist article is relevant when a meaning must cross languages; English itself remains one language among many. Readers working entirely between non-English languages should use the same invariants without making English a mandatory pivot.
For authoritative electricity-system vocabulary and data-model meaning, consult the official IEC references above and the versioned documentation of the relevant utility or system supplier. An educational explanation is not a license to perform field operations. A public service notice must always be reconciled with the issuing operator’s current information.
Closing principle: preserve the electrical reality, not merely the electrical words
The best translation of a power-system record is not the most impressive-looking technical prose. It is the version that preserves the intended audience’s understanding of the same equipment, state, evidence, quantity, event and authorisation. When the source records an alarm, translate the alarm rather than inventing the fault. When it reports a forecast, preserve the forecast rather than promising an outcome. When a device is identified by a code, protect that identity while making the public explanation readable. When a status is uncertain, keep it uncertain until an authorized source resolves it.
A reliable translator therefore behaves like a steward of meaning across boundaries. The language changes; the unit, event, object, confidence and provenance do not change without explicit authority. By combining concept-based terminology, claim mapping, numeric and time checks, domain review, accessible communication, version discipline and documented exceptions, multilingual utility information becomes more intelligible without becoming less trustworthy. That is the transferable art: not merely understanding a technical word, but preserving the conditions under which its meaning is true.