Master Art of Translation | The Semiconductor Fabrication, Wafer Testing & Chip Traceability Translation System — How Recipes, Equipment States, Yield Maps, Packaging and Reliability Records Keep Their Meaning

Master Art of Translation · Specialist world system · Child 520 · Original longform field manual

Semiconductor manufacturing translation means translating wafer-fab procedures, chip production records, process recipes, equipment alarms, wafer test reports, substrate maps and semiconductor packaging instructions without changing what engineers, operators and automated systems must do. This guide explains how technical translators preserve lot identity, recipe revision, measurement units, equipment states, yield evidence, safety restrictions and traceability when manufacturing information crosses languages. Its focus is translation quality, technical meaning and documented verification—not instructions for operating a fabrication plant.

People searching for semiconductor translation, wafer fabrication document translation, chip manufacturing terminology, SECS/GEM translation, wafer test report localization, semiconductor process recipe translation or wafer map translation need more than a list of equivalent words. The translator must distinguish a displayed description from an executable identifier, a wafer lot from an individual wafer, a technical alarm from its acknowledgment, a machine state from a production release, and a failing test from a decision to scrap material. A fluent sentence may still be operationally false.

The hidden problem is that a single translated word can silently change evidence or authorization: hold can become release, monitor can become control, qualified can become guaranteed, and coordinate origin can become an apparently harmless map label. This field manual turns those hazards into a repeatable method: identify the artifact and decision owner; separate immutable data from human-readable language; map exact technical conditions; test traceability, state transitions and units; then approve only a version whose recipient can make the same authorized decision as the source recipient.

This is a specialist child of Master Art of Translation | The Complete System for Moving Meaning Between Languages. It neither replaces that root nor takes ownership of the protected Vocabulary Learning Hub or How English Works | Translation and Meaning Across Languages. Those two routes explain lexical mastery and language interpretation; this chapter owns the narrowly defined industrial problem of translating semiconductor production evidence safely.

At a glance: scope, reader and outcome

The reader may be a professional translator, technical writer, manufacturing engineer, equipment supplier, document controller, test engineer, quality professional, customer reviewer, safety specialist, software localizer or auditor. No knowledge of a particular language pair is assumed. Worked examples use invented English-language source fragments and hypothetical defective glosses, deliberately avoiding fake foreign-language translations or real customer records. Their job is to reveal a decision boundary that must remain intact in any target language.

Search intent: find what semiconductor manufacturing translation really includes and how it differs from ordinary product-copy localization.
Hidden problem: machine identifiers, evidence, authority, time, units and state transitions do not behave like freely editable prose.
Mechanism: reconstruct the production object and the decision it supports, then protect each invariant.
Repeatable method: follow sixteen gates from brief and authority through review, regression tests and controlled release.
Failure modes: learn where plausible wording breaks recipe, mapping, safety, traceability, release or qualification meaning.
Worked examples: use fictional diagnostics from the front-end fab to wafer probe, packaging, test and downstream quality.
Transfer: apply the same reasoning to software strings, supplier documentation, training and independent audit.

1. What semiconductor manufacturing translation is—and is not

A semiconductor manufacturing document is rarely just an essay about technology. It may be a work instruction that defines conditions; a material specification that distinguishes one accepted input from a different grade; an engineering change record that makes a particular revision effective; a process record that reports what happened; a tool message that a host computer must parse; or a customer-facing statement that communicates a limited qualification. Translation is successful only when those functions remain distinguishable. The same string can appear in an operator screen, a purchasing contract and a historical audit trail, yet each occurrence may require a different treatment because its authority, audience and permitted edit surface differ.

In wafer fabrication, steps such as coating, patterning, etching, depositing films, implanting dopants, cleaning, thermal processing and polishing change the material. Metrology and inspection provide observations. Process control decides whether those observations trigger review or intervention. Equipment automation exchanges structured messages with factory systems. Wafer testing assigns results to devices; assembly and packaging preserve the link between devices and packages; qualification and reliability activities evaluate particular risks under defined plans. No translated document should imply that all these activities share one acceptance criterion, one vocabulary or one universal workflow.

Separate six objects before touching a source file. The physical object may be a wafer, die, package, carrier, tool or specimen. The information object may be a lot identifier, map, work instruction, data record or software field. The state may be planned, in process, complete, held, reviewed or released. The measurement has a unit, method, sampling context and sometimes uncertainty. The authority belongs to a named role or validated workflow, not to the translator. The revision establishes which terms and acceptance logic were active when a decision occurred. If any of these six objects is collapsed into another, translation becomes an undocumented process change.

A practical localization brief distinguishes free text that can be translated, controlled terms that need approved equivalents, keys or enumerations that must remain byte-for-byte identical, numeric evidence that must not be recalculated, and regulatory or contractual declarations that require authorized review. Operators benefit from clear language; machines benefit from stable protocols; auditors benefit from preserved records. Those are different beneficiaries. A release test therefore cannot be a single spell-check or a single score of fluency.

2. The hidden problem: a sentence can change an entire production decision

Imagine a traveler instruction saying that a lot is pending an engineering disposition after an out-of-control statistical signal. A hasty gloss might say that the lot has failed and must be scrapped. Another might say the issue was acknowledged and production can proceed. Both sound decisive, but neither is supported. The source only establishes a hold and the need for a separate authorized disposition. This is not a stylistic nicety: it is a chain-of-authority problem. Translate the current state, preserve the non-decision, identify who can resolve it and keep the material identity attached to the record.

Likewise, a wafer-map legend may contain entries that look like words but are actually coded values. A target-language interface may display a localized explanation of a bin, yet the internal bin code, map origin, axis orientation and association with a wafer must remain unchanged. A rotation error may move a quality decision from one die to another. At that point the translation is no longer about vocabulary. It concerns the referential integrity of a production system.

Ambiguity also hides inside apparently familiar quality terms. A pass on one electrical test does not necessarily mean a product is fully qualified. A capability observation is not the same as a specification limit; an action limit is not necessarily a physical device failure; a test retake is not automatically permitted; a maintenance acknowledgment does not imply corrective work is complete. These distinctions should be visible in bilingual glossaries and in the test cases used to authorize a release.

There is no universal plant-independent translation for every acronym, shorthand label, equipment state or alarm code. Many terms are manufacturer- or contract-specific. Publicly available SEMI, IEC and IPC/JEDEC information provides a terminology framework, but the controlled source procedure, licensed standard, supplier interface specification and site-approved glossary govern the actual job. When that evidence is absent, the correct action is to flag a query, not to invent an equivalence.

3. The mechanism: preserve fourteen invariants of industrial meaning

1. Entity identity. Keep lot ID, wafer ID, device ID, carrier ID, tool ID and customer part number attached to their correct objects. A readable label is not an authority to rename a key. This invariant should have a named source field, a permitted translation treatment and a verification owner; a beautiful sentence that fails the invariant cannot pass final quality review.

2. Revision identity. Maintain the effective source revision, applicability date, approval status and supersession relationship. Do not merge a proposed change with an approved procedure. This invariant should have a named source field, a permitted translation treatment and a verification owner; a beautiful sentence that fails the invariant cannot pass final quality review.

3. State semantics. Separate observation, request, acknowledgment, completion, hold, release and closure. A translated status verb must not perform a new state transition. This invariant should have a named source field, a permitted translation treatment and a verification owner; a beautiful sentence that fails the invariant cannot pass final quality review.

4. Responsibility. Represent who records, reviews, approves, executes and is notified. An impersonal construction must not hide a required independent approval. This invariant should have a named source field, a permitted translation treatment and a verification owner; a beautiful sentence that fails the invariant cannot pass final quality review.

5. Condition and exception. Protect if, only if, unless, not before, not after, pending, at least and no more than. Losing a condition can reverse an instruction. This invariant should have a named source field, a permitted translation treatment and a verification owner; a beautiful sentence that fails the invariant cannot pass final quality review.

6. Measurement identity. Carry quantity, unit, precision, method, sample and reference condition together; do not silently convert or round data in an evidentiary record. This invariant should have a named source field, a permitted translation treatment and a verification owner; a beautiful sentence that fails the invariant cannot pass final quality review.

7. Time identity. Distinguish wall-clock date, event time, duration, process elapsed time and audit timestamp. A timezone conversion requires authorization and an explicit basis. This invariant should have a named source field, a permitted translation treatment and a verification owner; a beautiful sentence that fails the invariant cannot pass final quality review.

8. Equipment boundary. Keep operator display strings separate from host commands, message IDs, stream/function identifiers, event IDs and variable names. This invariant should have a named source field, a permitted translation treatment and a verification owner; a beautiful sentence that fails the invariant cannot pass final quality review.

9. Map geometry. Protect wafer face, notch orientation, map origin, coordinate direction, sampling pattern, bin association and die identity. This invariant should have a named source field, a permitted translation treatment and a verification owner; a beautiful sentence that fails the invariant cannot pass final quality review.

10. Test meaning. Separate test condition, result, limit, test program revision, retest authorization, bin assignment and final disposition. This invariant should have a named source field, a permitted translation treatment and a verification owner; a beautiful sentence that fails the invariant cannot pass final quality review.

11. Safety meaning. Keep warnings, interlock states, hazardous-energy controls, restricted access, protective measures and emergency roles explicit and conservative. This invariant should have a named source field, a permitted translation treatment and a verification owner; a beautiful sentence that fails the invariant cannot pass final quality review.

12. Quality evidence. Distinguish an observation, deviation, nonconformance, containment measure, root cause, corrective action and effectiveness check. This invariant should have a named source field, a permitted translation treatment and a verification owner; a beautiful sentence that fails the invariant cannot pass final quality review.

13. Security authority. Do not turn technical vulnerability notices into permission to bypass security controls or imply a revision is approved when it is only proposed. This invariant should have a named source field, a permitted translation treatment and a verification owner; a beautiful sentence that fails the invariant cannot pass final quality review.

14. Recipient decision. An expert reading the target language must be able to make the same authorized operational decision as an expert reading the source, given the same evidence. This invariant should have a named source field, a permitted translation treatment and a verification owner; a beautiful sentence that fails the invariant cannot pass final quality review.

Think of the translation as a view onto a controlled manufacturing state rather than a replacement for that state. Human words may change by locale; identifiers and facts remain anchored. A bilingual rendering may therefore include an English technical term in parentheses, a localized explanatory label and an unchanged symbolic code. This is deliberate redundancy when it prevents category errors. Do not replace a supplier’s immutable label merely because an alternative word is more elegant.

4. The repeatable sixteen-gate translation and release method

Gate 01 — Intake and intended use. Record the artifact type, language pair, audience, manufacturing stage, software or document release and reason for translation. Identify whether it guides action, documents evidence or explains a result. Evidence of completion should be attached to the translation job, because a checklist tick without a checked artifact gives no reliable account of why the result is safe to use. For a high-consequence document, unresolved questions remain visible and block release until a qualified owner accepts or resolves them.

Gate 02 — Authority map. Identify the controlling procedure, supplier document, license, regulatory requirements and subject-matter approver. An outdated unofficial glossary is not a controlling authority. Evidence of completion should be attached to the translation job, because a checklist tick without a checked artifact gives no reliable account of why the result is safe to use. For a high-consequence document, unresolved questions remain visible and block release until a qualified owner accepts or resolves them.

Gate 03 — Object graph. List entities—lots, wafers, tools, recipes, carriers, devices, employees, systems—and the edges between them. Mark where source text labels those relationships. Evidence of completion should be attached to the translation job, because a checklist tick without a checked artifact gives no reliable account of why the result is safe to use. For a high-consequence document, unresolved questions remain visible and block release until a qualified owner accepts or resolves them.

Gate 04 — Immutable-token inventory. Freeze IDs, message codes, version tags, enum values, structured field names, map axes, checksums, filenames and test parameters according to the interface contract. Evidence of completion should be attached to the translation job, because a checklist tick without a checked artifact gives no reliable account of why the result is safe to use. For a high-consequence document, unresolved questions remain visible and block release until a qualified owner accepts or resolves them.

Gate 05 — Terminology and homographs. Record preferred equivalents, unacceptable near-synonyms and context notes for terms such as lot, hold, route, yield, bin, alarm, acknowledge, purge, qualification and disposition. Evidence of completion should be attached to the translation job, because a checklist tick without a checked artifact gives no reliable account of why the result is safe to use. For a high-consequence document, unresolved questions remain visible and block release until a qualified owner accepts or resolves them.

Gate 06 — Condition extraction. Underline every negation, threshold, conditional conjunction, exception, temporal restriction and role restriction; compare them against the source decision table. Evidence of completion should be attached to the translation job, because a checklist tick without a checked artifact gives no reliable account of why the result is safe to use. For a high-consequence document, unresolved questions remain visible and block release until a qualified owner accepts or resolves them.

Gate 07 — Unit and number protection. Validate numeric fields, symbols, decimal separators, ranges, significant figures, temperatures, pressure units, concentrations, ratios and time bases without assuming a conversion is authorized. Evidence of completion should be attached to the translation job, because a checklist tick without a checked artifact gives no reliable account of why the result is safe to use. For a high-consequence document, unresolved questions remain visible and block release until a qualified owner accepts or resolves them.

Gate 08 — Safety triage. Escalate any safety-critical or equipment-control wording to the approved technical owner. Do not infer interlock bypass or hazardous-material response procedures. Evidence of completion should be attached to the translation job, because a checklist tick without a checked artifact gives no reliable account of why the result is safe to use. For a high-consequence document, unresolved questions remain visible and block release until a qualified owner accepts or resolves them.

Gate 09 — First translation pass. Translate prose naturally while retaining all invariants. Use a comment or question where source ambiguity cannot be resolved from evidence. Evidence of completion should be attached to the translation job, because a checklist tick without a checked artifact gives no reliable account of why the result is safe to use. For a high-consequence document, unresolved questions remain visible and block release until a qualified owner accepts or resolves them.

Gate 10 — Independent technical review. Ask a domain reviewer to compare intended actions and constraints, not just grammatical accuracy. Document their decision and unresolved exceptions. Evidence of completion should be attached to the translation job, because a checklist tick without a checked artifact gives no reliable account of why the result is safe to use. For a high-consequence document, unresolved questions remain visible and block release until a qualified owner accepts or resolves them.

Gate 11 — Machine boundary review. Test CSV, XML, JSON, MES, recipe and SECS/GEM surfaces for accidental key translation, punctuation changes, encoding damage or schema drift. Evidence of completion should be attached to the translation job, because a checklist tick without a checked artifact gives no reliable account of why the result is safe to use. For a high-consequence document, unresolved questions remain visible and block release until a qualified owner accepts or resolves them.

Gate 12 — Bidirectional map and data check. Where coordinates or substrate maps are involved, compare references, orientation, origin, associations and record counts with the original. Evidence of completion should be attached to the translation job, because a checklist tick without a checked artifact gives no reliable account of why the result is safe to use. For a high-consequence document, unresolved questions remain visible and block release until a qualified owner accepts or resolves them.

Gate 13 — Human-interface test. Check truncation, warnings, button labels, accessibility text, screen order, status color dependence and recovery language using realistic operator tasks. Evidence of completion should be attached to the translation job, because a checklist tick without a checked artifact gives no reliable account of why the result is safe to use. For a high-consequence document, unresolved questions remain visible and block release until a qualified owner accepts or resolves them.

Gate 14 — Adversarial diagnostic. Try to force a false release, misplaced die, concealed safety warning, changed measurement or invalid permission using the localized text. Record whether the attempt fails. Evidence of completion should be attached to the translation job, because a checklist tick without a checked artifact gives no reliable account of why the result is safe to use. For a high-consequence document, unresolved questions remain visible and block release until a qualified owner accepts or resolves them.

Gate 15 — Approval and version receipt. Keep source version, target language, glossary version, test evidence, reviewer identity, date and approval status. Publish only the approved target artifact. Evidence of completion should be attached to the translation job, because a checklist tick without a checked artifact gives no reliable account of why the result is safe to use. For a high-consequence document, unresolved questions remain visible and block release until a qualified owner accepts or resolves them.

Gate 16 — Change and correction control. After any source change, determine affected units, rerun relevant tests and preserve the earlier release record. Corrections should be visible rather than silently overwritten. Evidence of completion should be attached to the translation job, because a checklist tick without a checked artifact gives no reliable account of why the result is safe to use. For a high-consequence document, unresolved questions remain visible and block release until a qualified owner accepts or resolves them.

5. Failure modes: attractive language, broken evidence

Failure 1: Hold turned into reject. A temporary material control becomes permanent disposition. Repair by restoring the current state and the separate disposition authority. A reviewer can test this by asking which downstream decision becomes possible after reading the target that was impossible in the source. If the target creates new permission, deletes a prerequisite or changes an object identity, the translation should fail even if its vocabulary is grammatical.

Failure 2: Acknowledge turned into clear. An alarm-confirmation action is confused with fault removal. Keep the event log, equipment state and corrective task distinct. A reviewer can test this by asking which downstream decision becomes possible after reading the target that was impossible in the source. If the target creates new permission, deletes a prerequisite or changes an object identity, the translation should fail even if its vocabulary is grammatical.

Failure 3: Recipe display name substituted for recipe ID. A localized string is pasted into a machine field. Maintain separate display and canonical identifier fields. A reviewer can test this by asking which downstream decision becomes possible after reading the target that was impossible in the source. If the target creates new permission, deletes a prerequisite or changes an object identity, the translation should fail even if its vocabulary is grammatical.

Failure 4: Wrong wafer side or notch orientation. A map is mirrored or rotated while labels appear correct. Verify physical reference conventions before approving a legend. A reviewer can test this by asking which downstream decision becomes possible after reading the target that was impossible in the source. If the target creates new permission, deletes a prerequisite or changes an object identity, the translation should fail even if its vocabulary is grammatical.

Failure 5: Soft bin mistaken for physical bin. A logical test classification is treated as a package sort destination. Read the specific test-program mapping. A reviewer can test this by asking which downstream decision becomes possible after reading the target that was impossible in the source. If the target creates new permission, deletes a prerequisite or changes an object identity, the translation should fail even if its vocabulary is grammatical.

Failure 6: Process limit mistaken for control limit. A statistical surveillance boundary becomes a product specification. State the owner and statistical basis of each threshold. A reviewer can test this by asking which downstream decision becomes possible after reading the target that was impossible in the source. If the target creates new permission, deletes a prerequisite or changes an object identity, the translation should fail even if its vocabulary is grammatical.

Failure 7: Minimum mistaken for target. A compliance floor becomes a recommended operating setpoint. Preserve constraint modality. A reviewer can test this by asking which downstream decision becomes possible after reading the target that was impossible in the source. If the target creates new permission, deletes a prerequisite or changes an object identity, the translation should fail even if its vocabulary is grammatical.

Failure 8: Exception silently removed. A sentence loses an unless clause, allowing the wrong material or recipe. Test the boundary case, not just the normal case. A reviewer can test this by asking which downstream decision becomes possible after reading the target that was impossible in the source. If the target creates new permission, deletes a prerequisite or changes an object identity, the translation should fail even if its vocabulary is grammatical.

Failure 9: Revision proposed becomes effective. A ballot, draft, deviation request or engineering plan is described as already in force. Keep the status and effective date. A reviewer can test this by asking which downstream decision becomes possible after reading the target that was impossible in the source. If the target creates new permission, deletes a prerequisite or changes an object identity, the translation should fail even if its vocabulary is grammatical.

Failure 10: Yield percentage treated as acceptance. An aggregate calculation is confused with a formal shipment decision. Keep denominator, population and customer disposition separate. A reviewer can test this by asking which downstream decision becomes possible after reading the target that was impossible in the source. If the target creates new permission, deletes a prerequisite or changes an object identity, the translation should fail even if its vocabulary is grammatical.

Failure 11: Measurement uncertainty erased. A reading with method limits becomes an exact guaranteed value. Preserve qualification and context. A reviewer can test this by asking which downstream decision becomes possible after reading the target that was impossible in the source. If the target creates new permission, deletes a prerequisite or changes an object identity, the translation should fail even if its vocabulary is grammatical.

Failure 12: Relative time becomes calendar time. Process duration and event timestamp are confused. Record timebase and zone in the target display. A reviewer can test this by asking which downstream decision becomes possible after reading the target that was impossible in the source. If the target creates new permission, deletes a prerequisite or changes an object identity, the translation should fail even if its vocabulary is grammatical.

Failure 13: Safety category flattened. Warning, caution, information and prohibited action collapse into generic advice. Reassess severity and required action. A reviewer can test this by asking which downstream decision becomes possible after reading the target that was impossible in the source. If the target creates new permission, deletes a prerequisite or changes an object identity, the translation should fail even if its vocabulary is grammatical.

Failure 14: Blank treated as zero. Missing results or not-tested values become numeric outcomes. Distinguish absent, zero and not applicable. A reviewer can test this by asking which downstream decision becomes possible after reading the target that was impossible in the source. If the target creates new permission, deletes a prerequisite or changes an object identity, the translation should fail even if its vocabulary is grammatical.

Failure 15: ESD terminology used as a guarantee. A control program label becomes a claim of no electrostatic risk. Preserve scope, limits and exceptions. A reviewer can test this by asking which downstream decision becomes possible after reading the target that was impossible in the source. If the target creates new permission, deletes a prerequisite or changes an object identity, the translation should fail even if its vocabulary is grammatical.

Failure 16: MSL classification treated as lifetime reliability. Moisture/reflow handling classification becomes a broad product durability promise. Keep classification separate from reliability qualification. A reviewer can test this by asking which downstream decision becomes possible after reading the target that was impossible in the source. If the target creates new permission, deletes a prerequisite or changes an object identity, the translation should fail even if its vocabulary is grammatical.

Failure 17: Protocol code localized. SECS-II message fields, parameter IDs or enum literals change with language. Lock machine-facing syntax and translate only approved display messages. A reviewer can test this by asking which downstream decision becomes possible after reading the target that was impossible in the source. If the target creates new permission, deletes a prerequisite or changes an object identity, the translation should fail even if its vocabulary is grammatical.

Failure 18: Carrier slot renumbered. A visually reordered list becomes a changed physical slot assignment. Test carrier and lot reconciliation. A reviewer can test this by asking which downstream decision becomes possible after reading the target that was impossible in the source. If the target creates new permission, deletes a prerequisite or changes an object identity, the translation should fail even if its vocabulary is grammatical.

Failure 19: Security advisory interpreted as authorization. A vulnerability or patch note turns into an instruction to disable controls. Refer to approved change and security procedures. A reviewer can test this by asking which downstream decision becomes possible after reading the target that was impossible in the source. If the target creates new permission, deletes a prerequisite or changes an object identity, the translation should fail even if its vocabulary is grammatical.

Failure 20: Customer notification overclaims closure. Containment or investigation is labeled complete resolution. Show what is done, open and pending. A reviewer can test this by asking which downstream decision becomes possible after reading the target that was impossible in the source. If the target creates new permission, deletes a prerequisite or changes an object identity, the translation should fail even if its vocabulary is grammatical.

Failure 21: Training example read as plant instruction. Illustrative values are mistaken for validated production parameters. Mark training data as fictional and consult controlled work instructions. A reviewer can test this by asking which downstream decision becomes possible after reading the target that was impossible in the source. If the target creates new permission, deletes a prerequisite or changes an object identity, the translation should fail even if its vocabulary is grammatical.

Failure 22: Acceptance evidence conflated. Supplier qualification, calibration status, validation and release authorization are described as if interchangeable. Keep their separate evidentiary roles. A reviewer can test this by asking which downstream decision becomes possible after reading the target that was impossible in the source. If the target creates new permission, deletes a prerequisite or changes an object identity, the translation should fail even if its vocabulary is grammatical.

6. The field laboratory: fictional worked diagnostics across the chip production chain

Every case below is invented for training. The text is not a semiconductor tool recipe, a certified safety procedure or an instruction to manipulate plant equipment. Values and IDs are illustrations. The goal is to train a disciplined reading method: identify what the source actually asserts, diagnose the dangerous near-synonym, write a controlled interpretation and design a test that would reveal a false equivalence before publication.

1. Document control, revisions and the job traveler

This domain concerns a controlled instruction linked to an effective revision and a named workflow. Its characteristic translation risk is that a proposed or obsolete instruction may masquerade as the current authority. The correct technical decision owner is document control and the process owner, not the language editor. Before a translator selects a preferred equivalent, identify the controlled object, the source revision, the permitted human-readable display surface and any machine-readable value that must remain immutable. A single bilingual terminology note should include an unacceptable near-synonym, because knowing what a term must not mean is often more useful than memorizing its dictionary definition.

A useful review exercise here is to trace one value from the first source record to its downstream recipient. Record whether the value is an instruction, an observation, a request, a classification or an approval. Then ask whether the target reader would preserve that category without access to the source language. If the translation changes the category, stop it before release, even if the surrounding sentence reads elegantly.

Diagnostic 001 — Effective revision

Fictional source situation. Traveler PT-84 references work instruction WI-17 rev C for operation 420; rev D is approved for a later effective date. This is part of a controlled instruction linked to an effective revision and a named workflow. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Treat rev D as current because its letter is later. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Check the referent before checking the adjective. An identifier can refer to a wafer, a device, a carrier, a recipe or a document revision; choosing the wrong object means that even technically familiar vocabulary will describe the wrong evidence. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Keep C as effective for the specified traveler until the documented change boundary. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Compare timestamps, approval receipts, revision applicability and traveler history. The release reviewer should record which invariant was challenged, what source artifact resolved the doubt and which target-language line was approved. A ‘looks good’ response is not a suitable acceptance record for a controlled production context. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Transfer question. What would happen if the same wording appeared not in a training example but in a live MES screen, equipment supplier manual, customer certificate or audit report? Locate the additional review authority needed in that context. If the relevant authority cannot be identified, preserve the original object’s identity and mark the issue unresolved rather than claiming that translation alone authorizes the change.

Diagnostic 002 — Training copy

Fictional source situation. An excerpt is marked ‘training reference only; not for execution’ and uses hypothetical parameter values. This is part of a controlled instruction linked to an effective revision and a named workflow. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Render the label as an ordinary job instruction. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Read the sentence against its state transition. A request, observation, acknowledgment, execution and final approval are different events. A translation that jumps forward in that chain grants an action the source did not authorize. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Retain its educational status conspicuously and block any implication that values are validated. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Check watermark, document class and access controls in both languages. Keep the translated explanation outside executable keys or codes unless the actual schema authorizes a change. This separation lets human readers learn the meaning without making a localization update look like a manufacturing software revision. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 003 — Superseded appendix

Fictional source situation. The appendix survives for historical evidence but no longer sets the inspection sampling plan. This is part of a controlled instruction linked to an effective revision and a named workflow. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Translate as a present inspection requirement. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Consider the measurement envelope. A number belongs to a quantity, unit, sampling population and method. Removing one of those may create an apparent fact that was never measured, even if every digit survives intact. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Label the appendix historical and identify the current controlling sampling procedure. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Inspect revision matrix, cross-references and change summary. Inspect one ordinary case and one boundary case. The boundary case is often more revealing because it exposes a missing ‘only’, an unearned approval or an omitted exception that fluent target-language readers would not otherwise notice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 004 — Unsigned change

Fictional source situation. A suggested tolerance adjustment is awaiting engineering and quality approval. This is part of a controlled instruction linked to an effective revision and a named workflow. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Say that engineering already changed the tolerance. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Use an adversarial counterexample: imagine a reader who is motivated to release material, bypass a review or claim compliance. If the translation can reasonably be read as authorizing that behavior while the source cannot, it needs revision. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Translate the proposal’s unapproved status and make no numerical substitution. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Find the complete sign-off chain before a target-language release. Treat unresolved ambiguity as an explicit query to the owner. Translators can explain uncertainty, but they cannot create an engineering disposition, a supplier guarantee or a safety clearance through language choice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 005 — Instruction vs record

Fictional source situation. A work instruction says what must be checked; the completed traveler records what was actually checked. This is part of a controlled instruction linked to an effective revision and a named workflow. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Convert record statements into future commands. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Ask first what the source proves and what it deliberately leaves undecided. A translation should never fill that gap with optimism or with a harsher conclusion. This distinction matters when the next recipient must record a disposition rather than merely understand a paragraph. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Preserve tense, event status and actual observed results rather than standardizing prose. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Pair instruction and executed record and reconcile what was required and completed. Save an independent comparison of source and target versions along with the test evidence. When the original changes, the record identifies which assumptions must be reopened rather than silently carrying old meanings forward. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

2. Lot identity, split, merge and material genealogy

This domain concerns a material genealogy connecting incoming wafers, processing lots and downstream devices. Its characteristic translation risk is that a family label can erase the identity of the exact population reviewed. The correct technical decision owner is manufacturing execution and material quality, not the language editor. Before a translator selects a preferred equivalent, identify the controlled object, the source revision, the permitted human-readable display surface and any machine-readable value that must remain immutable. A single bilingual terminology note should include an unacceptable near-synonym, because knowing what a term must not mean is often more useful than memorizing its dictionary definition.

Diagnostic 006 — Parent-child lot

Fictional source situation. Lot L-208 was split into L-208A and L-208B after a process hold; only the B child awaits review. This is part of a material genealogy connecting incoming wafers, processing lots and downstream devices. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Refer to both children as held parent material. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Read the sentence against its state transition. A request, observation, acknowledgment, execution and final approval are different events. A translation that jumps forward in that chain grants an action the source did not authorize. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Name the affected child precisely without assigning its state to its sibling. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Reconcile MES genealogy and each child’s status independently. Keep the translated explanation outside executable keys or codes unless the actual schema authorizes a change. This separation lets human readers learn the meaning without making a localization update look like a manufacturing software revision. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 007 — Wafer omission

Fictional source situation. Wafer 07 was removed for analysis and the remaining lot contains 23 wafers. This is part of a material genealogy connecting incoming wafers, processing lots and downstream devices. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Call the removed wafer scrapped or imply 24 remain. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Consider the measurement envelope. A number belongs to a quantity, unit, sampling population and method. Removing one of those may create an apparent fact that was never measured, even if every digit survives intact. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Separate removal for analysis from scrap and preserve the current count. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Compare slot map, transfer history and analysis request. Inspect one ordinary case and one boundary case. The boundary case is often more revealing because it exposes a missing ‘only’, an unearned approval or an omitted exception that fluent target-language readers would not otherwise notice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 008 — Part number alias

Fictional source situation. A customer-facing catalog name points to a canonical device part number and suffix. This is part of a material genealogy connecting incoming wafers, processing lots and downstream devices. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Translate the suffix as descriptive wording. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Use an adversarial counterexample: imagine a reader who is motivated to release material, bypass a review or claim compliance. If the translation can reasonably be read as authorizing that behavior while the source cannot, it needs revision. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Preserve the exact part number and explain the marketing label in free text. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Check bill of materials and customer master record. Treat unresolved ambiguity as an explicit query to the owner. Translators can explain uncertainty, but they cannot create an engineering disposition, a supplier guarantee or a safety clearance through language choice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 009 — Rework genealogy

Fictional source situation. The traveler shows a rework route for two designated wafers, not the entire batch. This is part of a material genealogy connecting incoming wafers, processing lots and downstream devices. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Describe the whole lot as reprocessed. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Ask first what the source proves and what it deliberately leaves undecided. A translation should never fill that gap with optimism or with a harsher conclusion. This distinction matters when the next recipient must record a disposition rather than merely understand a paragraph. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Keep wafer-level scope and distinct process passes without resetting historical data. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Review route history and actual wafer IDs. Save an independent comparison of source and target versions along with the test evidence. When the original changes, the record identifies which assumptions must be reopened rather than silently carrying old meanings forward. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 010 — Mixed material

Fictional source situation. Two input lots share a process chamber but remain separate product genealogy trees. This is part of a material genealogy connecting incoming wafers, processing lots and downstream devices. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Treat shared chamber history as a physical lot merge. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Check the referent before checking the adjective. An identifier can refer to a wafer, a device, a carrier, a recipe or a document revision; choosing the wrong object means that even technically familiar vocabulary will describe the wrong evidence. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Preserve common equipment exposure without inventing a new parent lot. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Rebuild the object relationships from load and unload records. The release reviewer should record which invariant was challenged, what source artifact resolved the doubt and which target-language line was approved. A ‘looks good’ response is not a suitable acceptance record for a controlled production context. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

3. Carrier, FOUP and slot numbering

This domain concerns a carrier assignment relating physical slots to individual wafers. Its characteristic translation risk is that translated indexing can silently change where material is expected to reside. The correct technical decision owner is material control, equipment integration and automation, not the language editor. Before a translator selects a preferred equivalent, identify the controlled object, the source revision, the permitted human-readable display surface and any machine-readable value that must remain immutable. A single bilingual terminology note should include an unacceptable near-synonym, because knowing what a term must not mean is often more useful than memorizing its dictionary definition.

Diagnostic 011 — Zero-based display

Fictional source situation. A tool API reports slot index 0 while the operator panel calls the same location slot 1. This is part of a carrier assignment relating physical slots to individual wafers. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Rewrite the API index to match panel numbering. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Consider the measurement envelope. A number belongs to a quantity, unit, sampling population and method. Removing one of those may create an apparent fact that was never measured, even if every digit survives intact. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Keep numerical systems in their own fields and document the explicit mapping. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Exercise first and last slot mappings against the interface contract. Inspect one ordinary case and one boundary case. The boundary case is often more revealing because it exposes a missing ‘only’, an unearned approval or an omitted exception that fluent target-language readers would not otherwise notice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 012 — Empty slot

Fictional source situation. Carrier map has an empty position between two occupied positions. This is part of a carrier assignment relating physical slots to individual wafers. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Remove the blank row as formatting clutter. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Use an adversarial counterexample: imagine a reader who is motivated to release material, bypass a review or claim compliance. If the translation can reasonably be read as authorizing that behavior while the source cannot, it needs revision. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Retain the empty physical position and the correct following wafer associations. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Compare loaded carrier and translated visualization position by position. Treat unresolved ambiguity as an explicit query to the owner. Translators can explain uncertainty, but they cannot create an engineering disposition, a supplier guarantee or a safety clearance through language choice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 013 — Carrier exchange

Fictional source situation. A FOUP ID changed after transfer; wafer IDs and process history remained unchanged. This is part of a carrier assignment relating physical slots to individual wafers. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Describe carrier replacement as a material lot change. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Ask first what the source proves and what it deliberately leaves undecided. A translation should never fill that gap with optimism or with a harsher conclusion. This distinction matters when the next recipient must record a disposition rather than merely understand a paragraph. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Separate container identity from the material inside it. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Reconcile move event, carrier history and wafer list. Save an independent comparison of source and target versions along with the test evidence. When the original changes, the record identifies which assumptions must be reopened rather than silently carrying old meanings forward. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 014 — Restricted opening

Fictional source situation. A status message says carrier access is inhibited pending interlock verification. This is part of a carrier assignment relating physical slots to individual wafers. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Translate inhibited as access allowed after acknowledgment. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Check the referent before checking the adjective. An identifier can refer to a wafer, a device, a carrier, a recipe or a document revision; choosing the wrong object means that even technically familiar vocabulary will describe the wrong evidence. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Preserve the blocked state and the required interlock evidence. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Perform a role-based simulation without actuating real equipment. The release reviewer should record which invariant was challenged, what source artifact resolved the doubt and which target-language line was approved. A ‘looks good’ response is not a suitable acceptance record for a controlled production context. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 015 — Duplicate scan

Fictional source situation. A carrier ID scan appears twice in a trace because the first reading was retried. This is part of a carrier assignment relating physical slots to individual wafers. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Call the duplicate a second physical carrier arrival. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Read the sentence against its state transition. A request, observation, acknowledgment, execution and final approval are different events. A translation that jumps forward in that chain grants an action the source did not authorize. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Distinguish repeated observation from a new material movement event. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Compare event IDs, timestamps and physical handoff receipts. Keep the translated explanation outside executable keys or codes unless the actual schema authorizes a change. This separation lets human readers learn the meaning without making a localization update look like a manufacturing software revision. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

4. Lithography masks, layers and overlay

This domain concerns patterning documents that distinguish mask layers and alignment measurements. Its characteristic translation risk is that a label error may point to a different pattern or measurement than intended. The correct technical decision owner is lithography engineering and mask data control, not the language editor. Before a translator selects a preferred equivalent, identify the controlled object, the source revision, the permitted human-readable display surface and any machine-readable value that must remain immutable. A single bilingual terminology note should include an unacceptable near-synonym, because knowing what a term must not mean is often more useful than memorizing its dictionary definition.

Diagnostic 016 — Layer code

Fictional source situation. Mask layer M12 is described by the friendly name ‘contact cut’ in a displayed legend. This is part of patterning documents that distinguish mask layers and alignment measurements. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Translate M12 into a localized phrase inside the CAD key. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Use an adversarial counterexample: imagine a reader who is motivated to release material, bypass a review or claim compliance. If the translation can reasonably be read as authorizing that behavior while the source cannot, it needs revision. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Keep mask code immutable and localize only the descriptive legend. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Compare design database and mask set identifiers. Treat unresolved ambiguity as an explicit query to the owner. Translators can explain uncertainty, but they cannot create an engineering disposition, a supplier guarantee or a safety clearance through language choice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 017 — Overlay observation

Fictional source situation. The report lists overlay error relative to a specified reference mark. This is part of patterning documents that distinguish mask layers and alignment measurements. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Render error as a commanded alignment offset. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Ask first what the source proves and what it deliberately leaves undecided. A translation should never fill that gap with optimism or with a harsher conclusion. This distinction matters when the next recipient must record a disposition rather than merely understand a paragraph. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Distinguish measured residual from control input and preserve reference frame. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Check metrology file, sign convention and correction record. Save an independent comparison of source and target versions along with the test evidence. When the original changes, the record identifies which assumptions must be reopened rather than silently carrying old meanings forward. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 018 — Exposure dose note

Fictional source situation. A training sheet includes example units for exposure dose and is marked nonproduction. This is part of patterning documents that distinguish mask layers and alignment measurements. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Convert its example into a validated setpoint. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Check the referent before checking the adjective. An identifier can refer to a wafer, a device, a carrier, a recipe or a document revision; choosing the wrong object means that even technically familiar vocabulary will describe the wrong evidence. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Explain the measurement concept without presenting a process parameter as approved. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Verify watermark and source of any authorized numeric value. The release reviewer should record which invariant was challenged, what source artifact resolved the doubt and which target-language line was approved. A ‘looks good’ response is not a suitable acceptance record for a controlled production context. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 019 — Reticle orientation

Fictional source situation. A transfer sheet identifies the reticle orientation by controlled graphic and code. This is part of patterning documents that distinguish mask layers and alignment measurements. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Mirror the graphic to suit reading direction. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Read the sentence against its state transition. A request, observation, acknowledgment, execution and final approval are different events. A translation that jumps forward in that chain grants an action the source did not authorize. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Preserve physical orientation while adapting text around the illustration. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Compare registration marks and approved image assets. Keep the translated explanation outside executable keys or codes unless the actual schema authorizes a change. This separation lets human readers learn the meaning without making a localization update look like a manufacturing software revision. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 020 — Sampling map

Fictional source situation. The inspection sampling coordinates select locations on a wafer, not the entire field population. This is part of patterning documents that distinguish mask layers and alignment measurements. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Describe the sample as a full-wafer inspection. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Consider the measurement envelope. A number belongs to a quantity, unit, sampling population and method. Removing one of those may create an apparent fact that was never measured, even if every digit survives intact. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Retain sample selection, population and limitation of inference. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Count measured sites and check the sampling specification. Inspect one ordinary case and one boundary case. The boundary case is often more revealing because it exposes a missing ‘only’, an unearned approval or an omitted exception that fluent target-language readers would not otherwise notice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

5. Etch, plasma endpoints and chamber conditions

This domain concerns process narratives distinguishing etch behavior, chamber state and observation. Its characteristic translation risk is that a descriptive outcome can be mistaken for an instruction to alter a controlled recipe. The correct technical decision owner is etch process engineering and equipment safety, not the language editor. Before a translator selects a preferred equivalent, identify the controlled object, the source revision, the permitted human-readable display surface and any machine-readable value that must remain immutable. A single bilingual terminology note should include an unacceptable near-synonym, because knowing what a term must not mean is often more useful than memorizing its dictionary definition.

Diagnostic 021 — Endpoint reached

Fictional source situation. A log marks that optical endpoint detection completed during a recorded process. This is part of process narratives distinguishing etch behavior, chamber state and observation. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Translate the event as a command to start endpoint detection. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Ask first what the source proves and what it deliberately leaves undecided. A translation should never fill that gap with optimism or with a harsher conclusion. This distinction matters when the next recipient must record a disposition rather than merely understand a paragraph. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Maintain the past event and identify the detector as the reporting mechanism. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Cross-check recipe trace and event timestamp. Save an independent comparison of source and target versions along with the test evidence. When the original changes, the record identifies which assumptions must be reopened rather than silently carrying old meanings forward. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 022 — Overetch duration

Fictional source situation. A process summary distinguishes main etch duration from a separately defined overetch segment. This is part of process narratives distinguishing etch behavior, chamber state and observation. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Combine both into a single editable time value. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Check the referent before checking the adjective. An identifier can refer to a wafer, a device, a carrier, a recipe or a document revision; choosing the wrong object means that even technically familiar vocabulary will describe the wrong evidence. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Keep segment identities and resist converting a report into a new operating instruction. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Review validated process flow and segment names. The release reviewer should record which invariant was challenged, what source artifact resolved the doubt and which target-language line was approved. A ‘looks good’ response is not a suitable acceptance record for a controlled production context. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 023 — Chamber conditioning

Fictional source situation. A maintenance record says chamber seasoning was completed before production qualification. This is part of process narratives distinguishing etch behavior, chamber state and observation. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Call seasoning sufficient proof of product release. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Read the sentence against its state transition. A request, observation, acknowledgment, execution and final approval are different events. A translation that jumps forward in that chain grants an action the source did not authorize. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Preserve the need for separate equipment or process qualification as applicable. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Check qualification signoff and actual chamber state. Keep the translated explanation outside executable keys or codes unless the actual schema authorizes a change. This separation lets human readers learn the meaning without making a localization update look like a manufacturing software revision. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 024 — Gas flow units

Fictional source situation. A report states a gas-flow reading in sccm with instrument and calibration context. This is part of process narratives distinguishing etch behavior, chamber state and observation. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Translate it as a pressure reading in pascals. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Consider the measurement envelope. A number belongs to a quantity, unit, sampling population and method. Removing one of those may create an apparent fact that was never measured, even if every digit survives intact. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Preserve measured quantity, unit and instrument rather than translating by association. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Verify field schema and traceable calibration reference. Inspect one ordinary case and one boundary case. The boundary case is often more revealing because it exposes a missing ‘only’, an unearned approval or an omitted exception that fluent target-language readers would not otherwise notice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 025 — Alarm observation

Fictional source situation. A plasma fault was detected and acknowledged, while the run disposition remains pending. This is part of process narratives distinguishing etch behavior, chamber state and observation. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Describe acknowledgment as successful completion of the run. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Use an adversarial counterexample: imagine a reader who is motivated to release material, bypass a review or claim compliance. If the translation can reasonably be read as authorizing that behavior while the source cannot, it needs revision. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Retain fault, acknowledgment and unresolved material disposition separately. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Compare alarm, control job and quality hold records. Treat unresolved ambiguity as an explicit query to the owner. Translators can explain uncertainty, but they cannot create an engineering disposition, a supplier guarantee or a safety clearance through language choice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

6. Deposition, films and targets

This domain concerns a film-build record linking intended layer, method and measured outcome. Its characteristic translation risk is that process-target language is confused with an actual measurement or qualification. The correct technical decision owner is thin-film process engineering and metrology, not the language editor. Before a translator selects a preferred equivalent, identify the controlled object, the source revision, the permitted human-readable display surface and any machine-readable value that must remain immutable. A single bilingual terminology note should include an unacceptable near-synonym, because knowing what a term must not mean is often more useful than memorizing its dictionary definition.

Diagnostic 026 — Thickness target

Fictional source situation. The recipe lists a nominal film-thickness objective distinct from post-process measurements. This is part of a film-build record linking intended layer, method and measured outcome. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. State the target as a measured wafer value. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Check the referent before checking the adjective. An identifier can refer to a wafer, a device, a carrier, a recipe or a document revision; choosing the wrong object means that even technically familiar vocabulary will describe the wrong evidence. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Preserve planned versus observed semantics and measurement method. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Check recipe field name and metrology report. The release reviewer should record which invariant was challenged, what source artifact resolved the doubt and which target-language line was approved. A ‘looks good’ response is not a suitable acceptance record for a controlled production context. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 027 — Composition ratio

Fictional source situation. A characterization report gives an approximate material ratio with uncertainty. This is part of a film-build record linking intended layer, method and measured outcome. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Translate the ratio as exact chemical stoichiometry. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Read the sentence against its state transition. A request, observation, acknowledgment, execution and final approval are different events. A translation that jumps forward in that chain grants an action the source did not authorize. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Retain approximation, uncertainty and analysis context. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Check report units and laboratory method. Keep the translated explanation outside executable keys or codes unless the actual schema authorizes a change. This separation lets human readers learn the meaning without making a localization update look like a manufacturing software revision. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 028 — Target change

Fictional source situation. A sputter target was replaced; the equipment still requires its normal post-service checks. This is part of a film-build record linking intended layer, method and measured outcome. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Call replacement a completed production release. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Consider the measurement envelope. A number belongs to a quantity, unit, sampling population and method. Removing one of those may create an apparent fact that was never measured, even if every digit survives intact. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Record replacement as a maintenance event and keep release authority separate. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Inspect replacement log and subsequent checks. Inspect one ordinary case and one boundary case. The boundary case is often more revealing because it exposes a missing ‘only’, an unearned approval or an omitted exception that fluent target-language readers would not otherwise notice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 029 — Precursor status

Fictional source situation. A supplier note distinguishes a material batch certificate from its shelf-use approval. This is part of a film-build record linking intended layer, method and measured outcome. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Treat the certificate as automatically authorizing all use. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Use an adversarial counterexample: imagine a reader who is motivated to release material, bypass a review or claim compliance. If the translation can reasonably be read as authorizing that behavior while the source cannot, it needs revision. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Represent material identity, storage conditions and local acceptance separately. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Check receiving criteria and expiry or retest basis. Treat unresolved ambiguity as an explicit query to the owner. Translators can explain uncertainty, but they cannot create an engineering disposition, a supplier guarantee or a safety clearance through language choice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 030 — Film name

Fictional source situation. An internal acronym refers to one particular stack layer in a multi-layer structure. This is part of a film-build record linking intended layer, method and measured outcome. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Expand it to the nearest familiar material word. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Ask first what the source proves and what it deliberately leaves undecided. A translation should never fill that gap with optimism or with a harsher conclusion. This distinction matters when the next recipient must record a disposition rather than merely understand a paragraph. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Use the approved stack dictionary and retain the canonical layer code. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Review approved cross-section and bill of process. Save an independent comparison of source and target versions along with the test evidence. When the original changes, the record identifies which assumptions must be reopened rather than silently carrying old meanings forward. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

7. Ion implantation and dopant records

This domain concerns an implantation record connecting species, dose, energy and affected wafer population. Its characteristic translation risk is that similar-looking units and symbols can change the subject of a measurement. The correct technical decision owner is implant process engineering and data integrity, not the language editor. Before a translator selects a preferred equivalent, identify the controlled object, the source revision, the permitted human-readable display surface and any machine-readable value that must remain immutable. A single bilingual terminology note should include an unacceptable near-synonym, because knowing what a term must not mean is often more useful than memorizing its dictionary definition.

Diagnostic 031 — Species notation

Fictional source situation. A completed traveler names one implanted species by its chemical symbol and controlled recipe. This is part of an implantation record connecting species, dose, energy and affected wafer population. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Replace that symbol with a lookalike element name. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Read the sentence against its state transition. A request, observation, acknowledgment, execution and final approval are different events. A translation that jumps forward in that chain grants an action the source did not authorize. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Keep the chemical identity and recipe association exact; ask for a domain review. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Compare approved recipe and material specification. Keep the translated explanation outside executable keys or codes unless the actual schema authorizes a change. This separation lets human readers learn the meaning without making a localization update look like a manufacturing software revision. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 032 — Dose versus energy

Fictional source situation. The report gives dose and implantation energy in separate fields. This is part of an implantation record connecting species, dose, energy and affected wafer population. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Convert energy into a dose because both are numerical parameters. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Consider the measurement envelope. A number belongs to a quantity, unit, sampling population and method. Removing one of those may create an apparent fact that was never measured, even if every digit survives intact. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Protect quantities, units and distinct explanatory text. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Validate field schema, units and instrument source. Inspect one ordinary case and one boundary case. The boundary case is often more revealing because it exposes a missing ‘only’, an unearned approval or an omitted exception that fluent target-language readers would not otherwise notice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 033 — Tilt angle

Fictional source situation. A historical record includes angle conventions referenced to a particular tool frame. This is part of an implantation record connecting species, dose, energy and affected wafer population. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Render angle without its frame reference. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Use an adversarial counterexample: imagine a reader who is motivated to release material, bypass a review or claim compliance. If the translation can reasonably be read as authorizing that behavior while the source cannot, it needs revision. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Retain sign, reference orientation and measurement origin in the display. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Compare source reference diagram and configuration log. Treat unresolved ambiguity as an explicit query to the owner. Translators can explain uncertainty, but they cannot create an engineering disposition, a supplier guarantee or a safety clearance through language choice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 034 — Wafer subset

Fictional source situation. Only the monitor wafer underwent a special characterization step. This is part of an implantation record connecting species, dose, energy and affected wafer population. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Say every product wafer was characterized. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Ask first what the source proves and what it deliberately leaves undecided. A translation should never fill that gap with optimism or with a harsher conclusion. This distinction matters when the next recipient must record a disposition rather than merely understand a paragraph. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Keep monitor and product wafer populations separate. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Reconcile actual sample identifiers. Save an independent comparison of source and target versions along with the test evidence. When the original changes, the record identifies which assumptions must be reopened rather than silently carrying old meanings forward. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 035 — Postimplant step

Fictional source situation. The route describes a later thermal operation but does not report its completion. This is part of an implantation record connecting species, dose, energy and affected wafer population. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Translate a future requirement as already completed. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Check the referent before checking the adjective. An identifier can refer to a wafer, a device, a carrier, a recipe or a document revision; choosing the wrong object means that even technically familiar vocabulary will describe the wrong evidence. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Maintain conditional or prospective tense until an event receipt exists. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Cross-check subsequent work history. The release reviewer should record which invariant was challenged, what source artifact resolved the doubt and which target-language line was approved. A ‘looks good’ response is not a suitable acceptance record for a controlled production context. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

8. Thermal operations and timebase

This domain concerns a furnace or anneal record distinguishing ramps, dwell segments and actual observations. Its characteristic translation risk is that timing terms can turn a logged event into an unauthorized operating instruction. The correct technical decision owner is thermal process owner and equipment data reviewer, not the language editor. Before a translator selects a preferred equivalent, identify the controlled object, the source revision, the permitted human-readable display surface and any machine-readable value that must remain immutable. A single bilingual terminology note should include an unacceptable near-synonym, because knowing what a term must not mean is often more useful than memorizing its dictionary definition.

Diagnostic 036 — Ramp vs soak

Fictional source situation. A training diagram labels a ramp segment and a steady-temperature dwell segment. This is part of a furnace or anneal record distinguishing ramps, dwell segments and actual observations. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Translate both segments as ‘heating period’. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Consider the measurement envelope. A number belongs to a quantity, unit, sampling population and method. Removing one of those may create an apparent fact that was never measured, even if every digit survives intact. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Keep the function of each segment separate without publishing process settings. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Compare segment definitions and trajectory plot. Inspect one ordinary case and one boundary case. The boundary case is often more revealing because it exposes a missing ‘only’, an unearned approval or an omitted exception that fluent target-language readers would not otherwise notice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 037 — Elapsed time

Fictional source situation. A chart marks minutes since recipe start, not the local clock time. This is part of a furnace or anneal record distinguishing ramps, dwell segments and actual observations. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Render the axis as calendar timestamps. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Use an adversarial counterexample: imagine a reader who is motivated to release material, bypass a review or claim compliance. If the translation can reasonably be read as authorizing that behavior while the source cannot, it needs revision. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Label the elapsed-time basis and prevent fabricated start dates. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Check chart metadata and event log. Treat unresolved ambiguity as an explicit query to the owner. Translators can explain uncertainty, but they cannot create an engineering disposition, a supplier guarantee or a safety clearance through language choice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 038 — Monitor thermocouple

Fictional source situation. A report notes the temperature sensor’s indicated value and location. This is part of a furnace or anneal record distinguishing ramps, dwell segments and actual observations. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Present it as exact wafer temperature without qualification. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Ask first what the source proves and what it deliberately leaves undecided. A translation should never fill that gap with optimism or with a harsher conclusion. This distinction matters when the next recipient must record a disposition rather than merely understand a paragraph. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Retain sensor identity, location and proxy status. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Compare instrumentation plan and measurement context. Save an independent comparison of source and target versions along with the test evidence. When the original changes, the record identifies which assumptions must be reopened rather than silently carrying old meanings forward. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 039 — Unplanned interruption

Fictional source situation. The record documents a pause and subsequent authorized evaluation. This is part of a furnace or anneal record distinguishing ramps, dwell segments and actual observations. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Describe the run as uninterrupted and fully conforming. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Check the referent before checking the adjective. An identifier can refer to a wafer, a device, a carrier, a recipe or a document revision; choosing the wrong object means that even technically familiar vocabulary will describe the wrong evidence. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Keep interruption evidence and any pending assessment. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Review run trace and disposition ticket. The release reviewer should record which invariant was challenged, what source artifact resolved the doubt and which target-language line was approved. A ‘looks good’ response is not a suitable acceptance record for a controlled production context. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 040 — Furnace batch

Fictional source situation. A single equipment run contains wafers from more than one parent lot. This is part of a furnace or anneal record distinguishing ramps, dwell segments and actual observations. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Conflate equipment batch and legal material lot identity. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Read the sentence against its state transition. A request, observation, acknowledgment, execution and final approval are different events. A translation that jumps forward in that chain grants an action the source did not authorize. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Distinguish processing batch from genealogy of each lot. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Compare load list and production histories. Keep the translated explanation outside executable keys or codes unless the actual schema authorizes a change. This separation lets human readers learn the meaning without making a localization update look like a manufacturing software revision. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

9. Wet cleaning and chemical compatibility

This domain concerns cleaning records that distinguish chemical identity, contamination and safety controls. Its characteristic translation risk is that ordinary synonyms can erase hazards or material compatibility restrictions. The correct technical decision owner is wet process engineering and EHS, not the language editor. Before a translator selects a preferred equivalent, identify the controlled object, the source revision, the permitted human-readable display surface and any machine-readable value that must remain immutable. A single bilingual terminology note should include an unacceptable near-synonym, because knowing what a term must not mean is often more useful than memorizing its dictionary definition.

Diagnostic 041 — Solution identity

Fictional source situation. A work instruction refers to a controlled chemical mixture by internal code and revision. This is part of cleaning records that distinguish chemical identity, contamination and safety controls. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Expand the code into an assumed public chemical name. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Use an adversarial counterexample: imagine a reader who is motivated to release material, bypass a review or claim compliance. If the translation can reasonably be read as authorizing that behavior while the source cannot, it needs revision. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Keep the exact controlled code and ask the chemical owner for its approved description. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Review safety documentation and formula access permissions. Treat unresolved ambiguity as an explicit query to the owner. Translators can explain uncertainty, but they cannot create an engineering disposition, a supplier guarantee or a safety clearance through language choice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 042 — Bath age

Fictional source situation. A status display reports a bath lifetime limit separately from current elapsed use. This is part of cleaning records that distinguish chemical identity, contamination and safety controls. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Treat maximum lifetime as remaining time. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Ask first what the source proves and what it deliberately leaves undecided. A translation should never fill that gap with optimism or with a harsher conclusion. This distinction matters when the next recipient must record a disposition rather than merely understand a paragraph. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Distinguish cap, current use and calculated remainder if authorized. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Check time origin and instrument record. Save an independent comparison of source and target versions along with the test evidence. When the original changes, the record identifies which assumptions must be reopened rather than silently carrying old meanings forward. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 043 — Particle inspection

Fictional source situation. An inspection finds particles on a monitor sample after cleaning. This is part of cleaning records that distinguish chemical identity, contamination and safety controls. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. State the whole lot has failed specification. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Check the referent before checking the adjective. An identifier can refer to a wafer, a device, a carrier, a recipe or a document revision; choosing the wrong object means that even technically familiar vocabulary will describe the wrong evidence. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Report the observation and sampling boundary, then the separate review state. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Compare sample map and quality criteria. The release reviewer should record which invariant was challenged, what source artifact resolved the doubt and which target-language line was approved. A ‘looks good’ response is not a suitable acceptance record for a controlled production context. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 044 — Drain versus recirculate

Fictional source situation. An operator text distinguishes recirculation status from a drain operation. This is part of cleaning records that distinguish chemical identity, contamination and safety controls. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Use a generic verb such as ‘circulate’ for both. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Read the sentence against its state transition. A request, observation, acknowledgment, execution and final approval are different events. A translation that jumps forward in that chain grants an action the source did not authorize. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Maintain action type and safety interlocks, and route critical wording to EHS. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Use interface-state tests on an isolated training copy. Keep the translated explanation outside executable keys or codes unless the actual schema authorizes a change. This separation lets human readers learn the meaning without making a localization update look like a manufacturing software revision. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 045 — Chemical lot expiry

Fictional source situation. Material is within the supplier shelf period but pending local acceptance. This is part of cleaning records that distinguish chemical identity, contamination and safety controls. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Label it ready for unrestricted use. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Consider the measurement envelope. A number belongs to a quantity, unit, sampling population and method. Removing one of those may create an apparent fact that was never measured, even if every digit survives intact. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Represent both conditions and avoid creating approval. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Check receiving record and site authorization. Inspect one ordinary case and one boundary case. The boundary case is often more revealing because it exposes a missing ‘only’, an unearned approval or an omitted exception that fluent target-language readers would not otherwise notice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

10. Chemical mechanical polishing and consumables

This domain concerns a polish process record separating removal, endpoint and consumable condition. Its characteristic translation risk is that relative terms may be translated as absolute equipment instructions. The correct technical decision owner is CMP engineering and materials quality, not the language editor. Before a translator selects a preferred equivalent, identify the controlled object, the source revision, the permitted human-readable display surface and any machine-readable value that must remain immutable. A single bilingual terminology note should include an unacceptable near-synonym, because knowing what a term must not mean is often more useful than memorizing its dictionary definition.

Diagnostic 046 — Removal rate

Fictional source situation. A study reports measured removal rate for samples under stated conditions. This is part of a polish process record separating removal, endpoint and consumable condition. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. State the figure as a universal equipment throughput rate. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Ask first what the source proves and what it deliberately leaves undecided. A translation should never fill that gap with optimism or with a harsher conclusion. This distinction matters when the next recipient must record a disposition rather than merely understand a paragraph. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Preserve metric, sample, unit and studied conditions. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Check study protocol and input material. Save an independent comparison of source and target versions along with the test evidence. When the original changes, the record identifies which assumptions must be reopened rather than silently carrying old meanings forward. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 047 — Dishing vs erosion

Fictional source situation. A surface map distinguishes localized feature dishing from wider pattern-density erosion. This is part of a polish process record separating removal, endpoint and consumable condition. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Translate both into the same generic defect name. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Check the referent before checking the adjective. An identifier can refer to a wafer, a device, a carrier, a recipe or a document revision; choosing the wrong object means that even technically familiar vocabulary will describe the wrong evidence. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Retain different geometry and measurement mechanisms. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Verify cross-section illustrations and measurement method. The release reviewer should record which invariant was challenged, what source artifact resolved the doubt and which target-language line was approved. A ‘looks good’ response is not a suitable acceptance record for a controlled production context. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 048 — Pad lifetime

Fictional source situation. A maintenance note shows pad usage approaching a review threshold. This is part of a polish process record separating removal, endpoint and consumable condition. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Call it expired and mandate immediate replacement. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Read the sentence against its state transition. A request, observation, acknowledgment, execution and final approval are different events. A translation that jumps forward in that chain grants an action the source did not authorize. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Describe the observed usage and the applicable decision rule. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Compare defined threshold and maintenance authority. Keep the translated explanation outside executable keys or codes unless the actual schema authorizes a change. This separation lets human readers learn the meaning without making a localization update look like a manufacturing software revision. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 049 — Slurry batch

Fictional source situation. The system records a slurry lot change mid-study. This is part of a polish process record separating removal, endpoint and consumable condition. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Erase the change as an unimportant supplier detail. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Consider the measurement envelope. A number belongs to a quantity, unit, sampling population and method. Removing one of those may create an apparent fact that was never measured, even if every digit survives intact. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Preserve material lot and time interval to keep comparisons interpretable. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Inspect material genealogy and process timestamps. Inspect one ordinary case and one boundary case. The boundary case is often more revealing because it exposes a missing ‘only’, an unearned approval or an omitted exception that fluent target-language readers would not otherwise notice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 050 — Post-CMP clean

Fictional source situation. An inspection form records cleaning completion but not final particulate acceptance. This is part of a polish process record separating removal, endpoint and consumable condition. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Present the form as proof of full lot release. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Use an adversarial counterexample: imagine a reader who is motivated to release material, bypass a review or claim compliance. If the translation can reasonably be read as authorizing that behavior while the source cannot, it needs revision. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Retain the missing acceptance step and reviewer responsibility. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Reconcile clean record, particle results and material status. Treat unresolved ambiguity as an explicit query to the owner. Translators can explain uncertainty, but they cannot create an engineering disposition, a supplier guarantee or a safety clearance through language choice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

11. Metrology, inspection and uncertainty

This domain concerns measurement evidence with instruments, conditions, calibration and sample boundaries. Its characteristic translation risk is that a displayed number can become a false absolute fact. The correct technical decision owner is metrology laboratory and process quality, not the language editor. Before a translator selects a preferred equivalent, identify the controlled object, the source revision, the permitted human-readable display surface and any machine-readable value that must remain immutable. A single bilingual terminology note should include an unacceptable near-synonym, because knowing what a term must not mean is often more useful than memorizing its dictionary definition.

Diagnostic 051 — Nominal vs measured

Fictional source situation. A drawing states nominal film thickness while a report lists measured values. This is part of measurement evidence with instruments, conditions, calibration and sample boundaries. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Use measured wording for the drawing’s nominal dimension. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Check the referent before checking the adjective. An identifier can refer to a wafer, a device, a carrier, a recipe or a document revision; choosing the wrong object means that even technically familiar vocabulary will describe the wrong evidence. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Separate design intent from empirical observation. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Check linked evidence and sampling plan. The release reviewer should record which invariant was challenged, what source artifact resolved the doubt and which target-language line was approved. A ‘looks good’ response is not a suitable acceptance record for a controlled production context. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 052 — Measurement uncertainty

Fictional source situation. A thickness result carries a reported uncertainty with stated confidence basis. This is part of measurement evidence with instruments, conditions, calibration and sample boundaries. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Drop uncertainty for readability. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Read the sentence against its state transition. A request, observation, acknowledgment, execution and final approval are different events. A translation that jumps forward in that chain grants an action the source did not authorize. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Retain uncertainty statement and method unless its removal is authorized and explained. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Review report footnotes, units and calibration chain. Keep the translated explanation outside executable keys or codes unless the actual schema authorizes a change. This separation lets human readers learn the meaning without making a localization update look like a manufacturing software revision. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 053 — Out-of-calibration

Fictional source situation. An instrument was discovered outside its allowed calibration interval. This is part of measurement evidence with instruments, conditions, calibration and sample boundaries. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Say historical readings are automatically invalid. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Consider the measurement envelope. A number belongs to a quantity, unit, sampling population and method. Removing one of those may create an apparent fact that was never measured, even if every digit survives intact. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Preserve finding and the still-pending impact assessment. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Check affected date range and retrospective review outcome. Inspect one ordinary case and one boundary case. The boundary case is often more revealing because it exposes a missing ‘only’, an unearned approval or an omitted exception that fluent target-language readers would not otherwise notice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 054 — Resolution vs accuracy

Fictional source situation. A sensor display provides fine digits but instrument accuracy is separately limited. This is part of measurement evidence with instruments, conditions, calibration and sample boundaries. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Translate display resolution as guaranteed accuracy. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Use an adversarial counterexample: imagine a reader who is motivated to release material, bypass a review or claim compliance. If the translation can reasonably be read as authorizing that behavior while the source cannot, it needs revision. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Explain the two characteristics without inferring precision from digits. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Compare instrument specification and verification plan. Treat unresolved ambiguity as an explicit query to the owner. Translators can explain uncertainty, but they cannot create an engineering disposition, a supplier guarantee or a safety clearance through language choice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 055 — Sampled sites

Fictional source situation. Five test sites represent a specified sample, not a census of the wafer. This is part of measurement evidence with instruments, conditions, calibration and sample boundaries. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Say the wafer was measured at every location. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Ask first what the source proves and what it deliberately leaves undecided. A translation should never fill that gap with optimism or with a harsher conclusion. This distinction matters when the next recipient must record a disposition rather than merely understand a paragraph. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. State sampled design and limitation of inference. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Verify site list and report scope. Save an independent comparison of source and target versions along with the test evidence. When the original changes, the record identifies which assumptions must be reopened rather than silently carrying old meanings forward. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

12. Statistical process control and excursions

This domain concerns a statistical control record linking populations, limits, signals and formal decisions. Its characteristic translation risk is that statistical alerts are wrongly translated into product compliance decisions. The correct technical decision owner is SPC owner and quality engineering, not the language editor. Before a translator selects a preferred equivalent, identify the controlled object, the source revision, the permitted human-readable display surface and any machine-readable value that must remain immutable. A single bilingual terminology note should include an unacceptable near-synonym, because knowing what a term must not mean is often more useful than memorizing its dictionary definition.

Diagnostic 056 — Control vs spec

Fictional source situation. A control-chart point crosses a control limit but product specification is evaluated separately. This is part of a statistical control record linking populations, limits, signals and formal decisions. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Call the material out of product specification. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Read the sentence against its state transition. A request, observation, acknowledgment, execution and final approval are different events. A translation that jumps forward in that chain grants an action the source did not authorize. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Label an out-of-control statistical signal without asserting an unproven product defect. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Compare chart basis and disposition criteria. Keep the translated explanation outside executable keys or codes unless the actual schema authorizes a change. This separation lets human readers learn the meaning without making a localization update look like a manufacturing software revision. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 057 — Warning vs action

Fictional source situation. A warning-level trend is defined differently from a mandatory action threshold. This is part of a statistical control record linking populations, limits, signals and formal decisions. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Translate warning as stop-process command. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Consider the measurement envelope. A number belongs to a quantity, unit, sampling population and method. Removing one of those may create an apparent fact that was never measured, even if every digit survives intact. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Preserve escalation stages and authorized responses. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Check rule set and alert severity. Inspect one ordinary case and one boundary case. The boundary case is often more revealing because it exposes a missing ‘only’, an unearned approval or an omitted exception that fluent target-language readers would not otherwise notice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 058 — Historical recalculation

Fictional source situation. A control chart is recalculated for investigation using a different baseline. This is part of a statistical control record linking populations, limits, signals and formal decisions. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Replace original published limits without recording history. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Use an adversarial counterexample: imagine a reader who is motivated to release material, bypass a review or claim compliance. If the translation can reasonably be read as authorizing that behavior while the source cannot, it needs revision. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Keep both baseline versions and the purpose of recalculation. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Compare chart provenance and change approval. Treat unresolved ambiguity as an explicit query to the owner. Translators can explain uncertainty, but they cannot create an engineering disposition, a supplier guarantee or a safety clearance through language choice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 059 — Yield denominator

Fictional source situation. A yield percent is calculated from tested dies, excluding untested positions. This is part of a statistical control record linking populations, limits, signals and formal decisions. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Present it as percent of all designed die positions. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Ask first what the source proves and what it deliberately leaves undecided. A translation should never fill that gap with optimism or with a harsher conclusion. This distinction matters when the next recipient must record a disposition rather than merely understand a paragraph. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Declare numerator, denominator and excluded population. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Recompute from source counts without changing the source record. Save an independent comparison of source and target versions along with the test evidence. When the original changes, the record identifies which assumptions must be reopened rather than silently carrying old meanings forward. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 060 — Open excursion

Fictional source situation. Containment is complete but root-cause review is pending. This is part of a statistical control record linking populations, limits, signals and formal decisions. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. State the excursion is closed. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Check the referent before checking the adjective. An identifier can refer to a wafer, a device, a carrier, a recipe or a document revision; choosing the wrong object means that even technically familiar vocabulary will describe the wrong evidence. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Keep completed containment and unresolved investigation as separate statuses. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Review action tracker and signoff. The release reviewer should record which invariant was challenged, what source artifact resolved the doubt and which target-language line was approved. A ‘looks good’ response is not a suitable acceptance record for a controlled production context. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

13. Manufacturing execution routes and process disposition

This domain concerns a MES route with operation IDs, permitted transitions and evidence requirements. Its characteristic translation risk is that a translated command or menu option may imply a transition not approved by the workflow. The correct technical decision owner is MES administrator, process engineering and quality, not the language editor. Before a translator selects a preferred equivalent, identify the controlled object, the source revision, the permitted human-readable display surface and any machine-readable value that must remain immutable. A single bilingual terminology note should include an unacceptable near-synonym, because knowing what a term must not mean is often more useful than memorizing its dictionary definition.

Diagnostic 061 — Skip vs waive

Fictional source situation. A proposed skip is awaiting formal authorization; the route contains no approved waiver. This is part of a MES route with operation IDs, permitted transitions and evidence requirements. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Translate the request as executed waiver. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Consider the measurement envelope. A number belongs to a quantity, unit, sampling population and method. Removing one of those may create an apparent fact that was never measured, even if every digit survives intact. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Preserve request state and approval requirement. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Check route transaction and change ticket. Inspect one ordinary case and one boundary case. The boundary case is often more revealing because it exposes a missing ‘only’, an unearned approval or an omitted exception that fluent target-language readers would not otherwise notice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 062 — Operation count

Fictional source situation. An operator guide shows operation 320 as a routing code, not a quantity. This is part of a MES route with operation IDs, permitted transitions and evidence requirements. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Localize 320 using an ordinal-word convention. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Use an adversarial counterexample: imagine a reader who is motivated to release material, bypass a review or claim compliance. If the translation can reasonably be read as authorizing that behavior while the source cannot, it needs revision. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Retain operation identifier and localize only the explanatory words. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Compare linked route steps and MES keys. Treat unresolved ambiguity as an explicit query to the owner. Translators can explain uncertainty, but they cannot create an engineering disposition, a supplier guarantee or a safety clearance through language choice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 063 — Rework permitted

Fictional source situation. Rework is available for a specified failure class only after a review. This is part of a MES route with operation IDs, permitted transitions and evidence requirements. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. State that every failed wafer can be reworked. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Ask first what the source proves and what it deliberately leaves undecided. A translation should never fill that gap with optimism or with a harsher conclusion. This distinction matters when the next recipient must record a disposition rather than merely understand a paragraph. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Preserve eligibility and the independent disposition. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Check permissions, failure classification and evidence. Save an independent comparison of source and target versions along with the test evidence. When the original changes, the record identifies which assumptions must be reopened rather than silently carrying old meanings forward. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 064 — Material hold

Fictional source situation. A hold blocks forward movement but allows specified review activities. This is part of a MES route with operation IDs, permitted transitions and evidence requirements. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Translate hold as destruction or unrestricted continuation. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Check the referent before checking the adjective. An identifier can refer to a wafer, a device, a carrier, a recipe or a document revision; choosing the wrong object means that even technically familiar vocabulary will describe the wrong evidence. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Keep the boundary of blocked and permitted actions exact. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Test workflow on training states. The release reviewer should record which invariant was challenged, what source artifact resolved the doubt and which target-language line was approved. A ‘looks good’ response is not a suitable acceptance record for a controlled production context. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 065 — Exception history

Fictional source situation. A completed deviation remains associated with the production batch in the audit record. This is part of a MES route with operation IDs, permitted transitions and evidence requirements. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Delete it because the deviation is closed. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Read the sentence against its state transition. A request, observation, acknowledgment, execution and final approval are different events. A translation that jumps forward in that chain grants an action the source did not authorize. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Retain historical deviation and closure outcome. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Reconcile permanent record and current state. Keep the translated explanation outside executable keys or codes unless the actual schema authorizes a change. This separation lets human readers learn the meaning without making a localization update look like a manufacturing software revision. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

14. SECS-II, GEM and host-equipment communication

This domain concerns machine-interface terms whose public human labels coexist with structured protocol fields. Its characteristic translation risk is that a perfectly readable target string can break host–equipment interoperability. The correct technical decision owner is equipment interface and factory automation, not the language editor. Before a translator selects a preferred equivalent, identify the controlled object, the source revision, the permitted human-readable display surface and any machine-readable value that must remain immutable. A single bilingual terminology note should include an unacceptable near-synonym, because knowing what a term must not mean is often more useful than memorizing its dictionary definition.

Diagnostic 066 — Stream function

Fictional source situation. A manual names a SECS-II SxFy message family and gives a human explanation. This is part of machine-interface terms whose public human labels coexist with structured protocol fields. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Translate the SxFy identifier or reorder its characters. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Use an adversarial counterexample: imagine a reader who is motivated to release material, bypass a review or claim compliance. If the translation can reasonably be read as authorizing that behavior while the source cannot, it needs revision. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Retain exact message designator and localize only explanation. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Run a comparison against the supplier’s interface specification. Treat unresolved ambiguity as an explicit query to the owner. Translators can explain uncertainty, but they cannot create an engineering disposition, a supplier guarantee or a safety clearance through language choice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 067 — Event report

Fictional source situation. A GEM collection event reports that equipment observed a transition. This is part of machine-interface terms whose public human labels coexist with structured protocol fields. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Turn the event description into an operator command. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Ask first what the source proves and what it deliberately leaves undecided. A translation should never fill that gap with optimism or with a harsher conclusion. This distinction matters when the next recipient must record a disposition rather than merely understand a paragraph. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Preserve reporting semantics, actor and timestamp. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Check event, variable and report definitions. Save an independent comparison of source and target versions along with the test evidence. When the original changes, the record identifies which assumptions must be reopened rather than silently carrying old meanings forward. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 068 — Alarm acknowledgment

Fictional source situation. Host notification confirms receipt of a fault event but does not clear the physical fault. This is part of machine-interface terms whose public human labels coexist with structured protocol fields. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Call receipt ‘repair complete’. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Check the referent before checking the adjective. An identifier can refer to a wafer, a device, a carrier, a recipe or a document revision; choosing the wrong object means that even technically familiar vocabulary will describe the wrong evidence. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Keep acknowledgment, alarm active state and maintenance work separate. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Simulate an unresolved alarm and inspect UI wording. The release reviewer should record which invariant was challenged, what source artifact resolved the doubt and which target-language line was approved. A ‘looks good’ response is not a suitable acceptance record for a controlled production context. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 069 — Control state

Fictional source situation. A mode label distinguishes remote control permission from actual execution of a job. This is part of machine-interface terms whose public human labels coexist with structured protocol fields. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Translate remote as run automatically. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Read the sentence against its state transition. A request, observation, acknowledgment, execution and final approval are different events. A translation that jumps forward in that chain grants an action the source did not authorize. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Keep control authority distinct from command execution. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Test state chart and rejected command cases. Keep the translated explanation outside executable keys or codes unless the actual schema authorizes a change. This separation lets human readers learn the meaning without making a localization update look like a manufacturing software revision. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 070 — Data variable

Fictional source situation. A numeric equipment variable has a stable ID but a multilingual display label. This is part of machine-interface terms whose public human labels coexist with structured protocol fields. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Replace the ID in the communications payload with translated words. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Consider the measurement envelope. A number belongs to a quantity, unit, sampling population and method. Removing one of those may create an apparent fact that was never measured, even if every digit survives intact. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Freeze variable identity while supporting localized descriptions. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Validate schema and sample trace replay. Inspect one ordinary case and one boundary case. The boundary case is often more revealing because it exposes a missing ‘only’, an unearned approval or an omitted exception that fluent target-language readers would not otherwise notice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

15. Equipment uptime, alarms and maintenance

This domain concerns an equipment condition history connected to preventive work and release decisions. Its characteristic translation risk is that performance measurements can become fabricated approval or availability claims. The correct technical decision owner is equipment maintenance and operations, not the language editor. Before a translator selects a preferred equivalent, identify the controlled object, the source revision, the permitted human-readable display surface and any machine-readable value that must remain immutable. A single bilingual terminology note should include an unacceptable near-synonym, because knowing what a term must not mean is often more useful than memorizing its dictionary definition.

Diagnostic 071 — Downtime class

Fictional source situation. A dashboard distinguishes planned preventive-maintenance time from unplanned fault time. This is part of an equipment condition history connected to preventive work and release decisions. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Translate both as unexpected equipment failure. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Ask first what the source proves and what it deliberately leaves undecided. A translation should never fill that gap with optimism or with a harsher conclusion. This distinction matters when the next recipient must record a disposition rather than merely understand a paragraph. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Retain categories and their statistical impact. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Compare E10-based category definitions used by the facility. Save an independent comparison of source and target versions along with the test evidence. When the original changes, the record identifies which assumptions must be reopened rather than silently carrying old meanings forward. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 072 — MTBF vs MTTR

Fictional source situation. A report gives mean time between failures and mean time to repair in separate measures. This is part of an equipment condition history connected to preventive work and release decisions. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Use one generic ‘average repair time’ label. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Check the referent before checking the adjective. An identifier can refer to a wafer, a device, a carrier, a recipe or a document revision; choosing the wrong object means that even technically familiar vocabulary will describe the wrong evidence. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Preserve numerator, denominator and meaning of each metric. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Check metric specification and underlying log. The release reviewer should record which invariant was challenged, what source artifact resolved the doubt and which target-language line was approved. A ‘looks good’ response is not a suitable acceptance record for a controlled production context. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 073 — PM completion

Fictional source situation. A technician closes a maintenance checklist; engineering process qualification remains open. This is part of an equipment condition history connected to preventive work and release decisions. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Present completion as tool production release. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Read the sentence against its state transition. A request, observation, acknowledgment, execution and final approval are different events. A translation that jumps forward in that chain grants an action the source did not authorize. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Describe completed PM and outstanding qualification explicitly. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Reconcile PM record, equipment state and release signoff. Keep the translated explanation outside executable keys or codes unless the actual schema authorizes a change. This separation lets human readers learn the meaning without making a localization update look like a manufacturing software revision. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 074 — Spare part substitution

Fictional source situation. An alternative spare part is proposed but compatibility is not yet approved. This is part of an equipment condition history connected to preventive work and release decisions. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Translate proposal as installed approved replacement. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Consider the measurement envelope. A number belongs to a quantity, unit, sampling population and method. Removing one of those may create an apparent fact that was never measured, even if every digit survives intact. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Keep proposal, part identity and pending engineering evaluation. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Check approved parts list and actual service record. Inspect one ordinary case and one boundary case. The boundary case is often more revealing because it exposes a missing ‘only’, an unearned approval or an omitted exception that fluent target-language readers would not otherwise notice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 075 — Alarm reset

Fictional source situation. A fault indication was reset after review but its investigation remains open. This is part of an equipment condition history connected to preventive work and release decisions. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Call reset ‘root cause fixed’. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Use an adversarial counterexample: imagine a reader who is motivated to release material, bypass a review or claim compliance. If the translation can reasonably be read as authorizing that behavior while the source cannot, it needs revision. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Preserve reset status separately from cause and preventive action. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Trace event, maintenance ticket and investigation record. Treat unresolved ambiguity as an explicit query to the owner. Translators can explain uncertainty, but they cannot create an engineering disposition, a supplier guarantee or a safety clearance through language choice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

16. Automated material handling and transfers

This domain concerns a movement transaction linking carrier, station, destination and receipt. Its characteristic translation risk is that a status word may move material on paper without a real handoff. The correct technical decision owner is AMHS and material control teams, not the language editor. Before a translator selects a preferred equivalent, identify the controlled object, the source revision, the permitted human-readable display surface and any machine-readable value that must remain immutable. A single bilingual terminology note should include an unacceptable near-synonym, because knowing what a term must not mean is often more useful than memorizing its dictionary definition.

Diagnostic 076 — Requested vs confirmed

Fictional source situation. A move request was accepted by scheduling but no transfer completion exists. This is part of a movement transaction linking carrier, station, destination and receipt. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Say the carrier arrived at the destination. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Check the referent before checking the adjective. An identifier can refer to a wafer, a device, a carrier, a recipe or a document revision; choosing the wrong object means that even technically familiar vocabulary will describe the wrong evidence. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Keep planned, dispatched, physically arrived and received statuses distinct. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Reconcile move IDs and handoff acknowledgments. The release reviewer should record which invariant was challenged, what source artifact resolved the doubt and which target-language line was approved. A ‘looks good’ response is not a suitable acceptance record for a controlled production context. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 077 — Destination alias

Fictional source situation. A station has a display name and immutable equipment identifier. This is part of a movement transaction linking carrier, station, destination and receipt. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Translate the identifier as if it were a descriptive destination. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Read the sentence against its state transition. A request, observation, acknowledgment, execution and final approval are different events. A translation that jumps forward in that chain grants an action the source did not authorize. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Preserve the station code, localize only help text. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Check routing database and command message. Keep the translated explanation outside executable keys or codes unless the actual schema authorizes a change. This separation lets human readers learn the meaning without making a localization update look like a manufacturing software revision. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 078 — Lost scan

Fictional source situation. The tracking service reports an unreadable barcode, not lost physical material. This is part of a movement transaction linking carrier, station, destination and receipt. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Translate it as material missing. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Consider the measurement envelope. A number belongs to a quantity, unit, sampling population and method. Removing one of those may create an apparent fact that was never measured, even if every digit survives intact. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Describe identity-read failure and required reconciliation. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Check physical inspection and barcode event. Inspect one ordinary case and one boundary case. The boundary case is often more revealing because it exposes a missing ‘only’, an unearned approval or an omitted exception that fluent target-language readers would not otherwise notice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 079 — Transfer abort

Fictional source situation. A move is aborted and the carrier remains at the original station. This is part of a movement transaction linking carrier, station, destination and receipt. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Describe the abort as a completed return movement. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Use an adversarial counterexample: imagine a reader who is motivated to release material, bypass a review or claim compliance. If the translation can reasonably be read as authorizing that behavior while the source cannot, it needs revision. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Preserve origin, destination and current verified location. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Compare physical inventory and sequence of move events. Treat unresolved ambiguity as an explicit query to the owner. Translators can explain uncertainty, but they cannot create an engineering disposition, a supplier guarantee or a safety clearance through language choice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 080 — Handoff error

Fictional source situation. A receiving station rejects a transfer because its expected carrier ID differs. This is part of a movement transaction linking carrier, station, destination and receipt. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Describe this as a product quality reject. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Ask first what the source proves and what it deliberately leaves undecided. A translation should never fill that gap with optimism or with a harsher conclusion. This distinction matters when the next recipient must record a disposition rather than merely understand a paragraph. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Keep logistics mismatch separate from material-quality disposition. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Read station logs and lot record. Save an independent comparison of source and target versions along with the test evidence. When the original changes, the record identifies which assumptions must be reopened rather than silently carrying old meanings forward. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

17. Wafer probing, test programs and binning

This domain concerns a wafer electrical test record linking an individual die to program, condition and bin. Its characteristic translation risk is that test classifications are mistaken for unconditional product release. The correct technical decision owner is product test engineering and yield quality, not the language editor. Before a translator selects a preferred equivalent, identify the controlled object, the source revision, the permitted human-readable display surface and any machine-readable value that must remain immutable. A single bilingual terminology note should include an unacceptable near-synonym, because knowing what a term must not mean is often more useful than memorizing its dictionary definition.

Diagnostic 081 — Soft vs hard bin

Fictional source situation. A test program defines logical soft bins and a separate physical sort mapping. This is part of a wafer electrical test record linking an individual die to program, condition and bin. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Use one word for both categories. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Read the sentence against its state transition. A request, observation, acknowledgment, execution and final approval are different events. A translation that jumps forward in that chain grants an action the source did not authorize. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Name logical classification and physical routing distinctly. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Reconcile test-program mapping and handler configuration. Keep the translated explanation outside executable keys or codes unless the actual schema authorizes a change. This separation lets human readers learn the meaning without making a localization update look like a manufacturing software revision. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 082 — Retest status

Fictional source situation. A die has an initial failure and a later retest under an approved protocol. This is part of a wafer electrical test record linking an individual die to program, condition and bin. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Replace the first result with the retest result. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Consider the measurement envelope. A number belongs to a quantity, unit, sampling population and method. Removing one of those may create an apparent fact that was never measured, even if every digit survives intact. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Preserve both observations, retest authorization and final disposition rule. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Check complete history and bin policy. Inspect one ordinary case and one boundary case. The boundary case is often more revealing because it exposes a missing ‘only’, an unearned approval or an omitted exception that fluent target-language readers would not otherwise notice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 083 — Untested die

Fictional source situation. Map positions include untested, invalid, tested fail and tested pass states. This is part of a wafer electrical test record linking an individual die to program, condition and bin. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Render untested as failing electrical specification. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Use an adversarial counterexample: imagine a reader who is motivated to release material, bypass a review or claim compliance. If the translation can reasonably be read as authorizing that behavior while the source cannot, it needs revision. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Keep missing test evidence distinct from negative result. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Inspect map legend and raw test records. Treat unresolved ambiguity as an explicit query to the owner. Translators can explain uncertainty, but they cannot create an engineering disposition, a supplier guarantee or a safety clearance through language choice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 084 — Program revision

Fictional source situation. The same part was tested under two approved program revisions on different dates. This is part of a wafer electrical test record linking an individual die to program, condition and bin. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Combine results without identifying program version. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Ask first what the source proves and what it deliberately leaves undecided. A translation should never fill that gap with optimism or with a harsher conclusion. This distinction matters when the next recipient must record a disposition rather than merely understand a paragraph. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Retain revision and effective date for each result. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Review program checksum, lot history and change receipt. Save an independent comparison of source and target versions along with the test evidence. When the original changes, the record identifies which assumptions must be reopened rather than silently carrying old meanings forward. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 085 — Guardband

Fictional source situation. An internal guardband is tighter than the external product limit. This is part of a wafer electrical test record linking an individual die to program, condition and bin. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Translate guardband as the published customer specification. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Check the referent before checking the adjective. An identifier can refer to a wafer, a device, a carrier, a recipe or a document revision; choosing the wrong object means that even technically familiar vocabulary will describe the wrong evidence. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. State the internal screening boundary and its separate purpose. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Compare customer specification and program limits. The release reviewer should record which invariant was challenged, what source artifact resolved the doubt and which target-language line was approved. A ‘looks good’ response is not a suitable acceptance record for a controlled production context. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

18. Substrate maps and SEMI E142 geometry

This domain concerns wafer, frame, strip or tray maps that attach locations to individual devices. Its characteristic translation risk is that a translated legend can corrupt position, bin or identity without changing any visible prose. The correct technical decision owner is test integration, assembly and substrate-map owners, not the language editor. Before a translator selects a preferred equivalent, identify the controlled object, the source revision, the permitted human-readable display surface and any machine-readable value that must remain immutable. A single bilingual terminology note should include an unacceptable near-synonym, because knowing what a term must not mean is often more useful than memorizing its dictionary definition.

Diagnostic 086 — Coordinate origin

Fictional source situation. Prober coordinates use a different origin convention from the production substrate map. This is part of wafer, frame, strip or tray maps that attach locations to individual devices. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Simply copy X-Y numbers into the E142 map. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Consider the measurement envelope. A number belongs to a quantity, unit, sampling population and method. Removing one of those may create an apparent fact that was never measured, even if every digit survives intact. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Preserve source coordinate system and require an approved transformation. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Test a known corner and a noncentral die against both frames. Inspect one ordinary case and one boundary case. The boundary case is often more revealing because it exposes a missing ‘only’, an unearned approval or an omitted exception that fluent target-language readers would not otherwise notice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 087 — Rotation

Fictional source situation. A substrate map has an explicit orientation and top-side convention. This is part of wafer, frame, strip or tray maps that attach locations to individual devices. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Mirror or rotate the map for visual reading direction. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Use an adversarial counterexample: imagine a reader who is motivated to release material, bypass a review or claim compliance. If the translation can reasonably be read as authorizing that behavior while the source cannot, it needs revision. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Keep physical orientation and explain it in target-language captions. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Compare origin, axis direction and orientation fields. Treat unresolved ambiguity as an explicit query to the owner. Translators can explain uncertainty, but they cannot create an engineering disposition, a supplier guarantee or a safety clearance through language choice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 088 — Bin overlay

Fictional source situation. A bin-code map records electrical classification by device location. This is part of wafer, frame, strip or tray maps that attach locations to individual devices. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Replace codes with translated words inside the map data. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Ask first what the source proves and what it deliberately leaves undecided. A translation should never fill that gap with optimism or with a harsher conclusion. This distinction matters when the next recipient must record a disposition rather than merely understand a paragraph. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Preserve machine bin values and provide localized lookup labels. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Round-trip map through schema validation and test sample points. Save an independent comparison of source and target versions along with the test evidence. When the original changes, the record identifies which assumptions must be reopened rather than silently carrying old meanings forward. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 089 — Transfer mapping

Fictional source situation. A die transferred from wafer to tray retains a source-device association. This is part of wafer, frame, strip or tray maps that attach locations to individual devices. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Translate ‘from’ and ‘to’ as interchangeable labels. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Check the referent before checking the adjective. An identifier can refer to a wafer, a device, a carrier, a recipe or a document revision; choosing the wrong object means that even technically familiar vocabulary will describe the wrong evidence. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Preserve direction, reference device and destination position. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Reconstruct trace from wafer die to tray location. The release reviewer should record which invariant was challenged, what source artifact resolved the doubt and which target-language line was approved. A ‘looks good’ response is not a suitable acceptance record for a controlled production context. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 090 — Revision proposal

Fictional source situation. An article describes a proposed coordinate-mapping subordinate standard still under standards work. This is part of wafer, frame, strip or tray maps that attach locations to individual devices. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Call the proposed document an existing mandatory rule. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Read the sentence against its state transition. A request, observation, acknowledgment, execution and final approval are different events. A translation that jumps forward in that chain grants an action the source did not authorize. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Report its proposal status and distinguish it from the currently published reference. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Verify dated SEMI publication and current approved implementation profile. Keep the translated explanation outside executable keys or codes unless the actual schema authorizes a change. This separation lets human readers learn the meaning without making a localization update look like a manufacturing software revision. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

19. Die preparation, attach and bonding

This domain concerns assembly evidence connecting die, carrier, bonding operation and inspected outcomes. Its characteristic translation risk is that component genealogy can be replaced by an attractive but untrue assembly narrative. The correct technical decision owner is assembly engineering and process quality, not the language editor. Before a translator selects a preferred equivalent, identify the controlled object, the source revision, the permitted human-readable display surface and any machine-readable value that must remain immutable. A single bilingual terminology note should include an unacceptable near-synonym, because knowing what a term must not mean is often more useful than memorizing its dictionary definition.

Diagnostic 091 — Die attach lot

Fictional source situation. A die-attach batch contains devices from two wafer lots with separate pedigrees. This is part of assembly evidence connecting die, carrier, bonding operation and inspected outcomes. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Refer to the assembled batch as one original wafer lot. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Use an adversarial counterexample: imagine a reader who is motivated to release material, bypass a review or claim compliance. If the translation can reasonably be read as authorizing that behavior while the source cannot, it needs revision. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Preserve genealogy by device or unit through assembly. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Trace actual input lot and finished unit ID. Treat unresolved ambiguity as an explicit query to the owner. Translators can explain uncertainty, but they cannot create an engineering disposition, a supplier guarantee or a safety clearance through language choice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 092 — Wire-bond pull

Fictional source situation. A sample mechanical test reports results for selected bonds. This is part of assembly evidence connecting die, carrier, bonding operation and inspected outcomes. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Describe the report as verification of every bond on every unit. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Ask first what the source proves and what it deliberately leaves undecided. A translation should never fill that gap with optimism or with a harsher conclusion. This distinction matters when the next recipient must record a disposition rather than merely understand a paragraph. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Keep the sampling method, measured population and test basis. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Check sample IDs and inspection plan. Save an independent comparison of source and target versions along with the test evidence. When the original changes, the record identifies which assumptions must be reopened rather than silently carrying old meanings forward. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 093 — Flip-chip orientation

Fictional source situation. A drawing identifies bump-array orientation by a fiducial and controlled image. This is part of assembly evidence connecting die, carrier, bonding operation and inspected outcomes. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Reorient the figure for left-to-right reading without correcting metadata. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Check the referent before checking the adjective. An identifier can refer to a wafer, a device, a carrier, a recipe or a document revision; choosing the wrong object means that even technically familiar vocabulary will describe the wrong evidence. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Keep the device orientation invariant and localize captions around it. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Compare fiducials and assembly drawing revision. The release reviewer should record which invariant was challenged, what source artifact resolved the doubt and which target-language line was approved. A ‘looks good’ response is not a suitable acceptance record for a controlled production context. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 094 — Underfill cure

Fictional source situation. A record shows underfill dispense complete and a later cure step pending. This is part of assembly evidence connecting die, carrier, bonding operation and inspected outcomes. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Call the package fully cured. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Read the sentence against its state transition. A request, observation, acknowledgment, execution and final approval are different events. A translation that jumps forward in that chain grants an action the source did not authorize. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Maintain process-sequence status and separate acceptance evidence. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Reconcile timestamps and traveler steps. Keep the translated explanation outside executable keys or codes unless the actual schema authorizes a change. This separation lets human readers learn the meaning without making a localization update look like a manufacturing software revision. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 095 — Bond failure

Fictional source situation. An inspection marks a defect for review with no final root-cause finding. This is part of assembly evidence connecting die, carrier, bonding operation and inspected outcomes. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Translate it as proven contamination from a supplier. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Consider the measurement envelope. A number belongs to a quantity, unit, sampling population and method. Removing one of those may create an apparent fact that was never measured, even if every digit survives intact. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Retain observation without inventing causation or liability. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Compare failure analysis and disposition ticket. Inspect one ordinary case and one boundary case. The boundary case is often more revealing because it exposes a missing ‘only’, an unearned approval or an omitted exception that fluent target-language readers would not otherwise notice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

20. Packaging, moisture sensitivity and dry handling

This domain concerns package classification and shipping labels linked to moisture exposure and reflow handling. Its characteristic translation risk is that handling restrictions can become invalid reliability promises. The correct technical decision owner is packaging engineering and customer quality, not the language editor. Before a translator selects a preferred equivalent, identify the controlled object, the source revision, the permitted human-readable display surface and any machine-readable value that must remain immutable. A single bilingual terminology note should include an unacceptable near-synonym, because knowing what a term must not mean is often more useful than memorizing its dictionary definition.

Diagnostic 096 — MSL classification

Fictional source situation. A label states a moisture sensitivity level according to an identified test basis. This is part of package classification and shipping labels linked to moisture exposure and reflow handling. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Translate it as a warranty against all humidity damage. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Ask first what the source proves and what it deliberately leaves undecided. A translation should never fill that gap with optimism or with a harsher conclusion. This distinction matters when the next recipient must record a disposition rather than merely understand a paragraph. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Keep classification’s process-specific function and document reference. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Compare approved label and classification record. Save an independent comparison of source and target versions along with the test evidence. When the original changes, the record identifies which assumptions must be reopened rather than silently carrying old meanings forward. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 097 — Floor life

Fictional source situation. A handling note distinguishes permitted exposure duration from elapsed floor life already consumed. This is part of package classification and shipping labels linked to moisture exposure and reflow handling. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Describe total allowed exposure as time remaining. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Check the referent before checking the adjective. An identifier can refer to a wafer, a device, a carrier, a recipe or a document revision; choosing the wrong object means that even technically familiar vocabulary will describe the wrong evidence. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Preserve arithmetic fields and current storage conditions. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Check opening time, cumulative exposure log and applicable rules. The release reviewer should record which invariant was challenged, what source artifact resolved the doubt and which target-language line was approved. A ‘looks good’ response is not a suitable acceptance record for a controlled production context. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 098 — Dry pack

Fictional source situation. Packaging record indicates moisture-barrier bag sealing, not a passed reliability qualification. This is part of package classification and shipping labels linked to moisture exposure and reflow handling. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Call sealing an assurance of product lifetime. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Read the sentence against its state transition. A request, observation, acknowledgment, execution and final approval are different events. A translation that jumps forward in that chain grants an action the source did not authorize. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. State packaging status and its narrow protective purpose. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Inspect packaging checklist and independent qualification record. Keep the translated explanation outside executable keys or codes unless the actual schema authorizes a change. This separation lets human readers learn the meaning without making a localization update look like a manufacturing software revision. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 099 — Reflow conditions

Fictional source situation. A specification identifies component classification parameters distinct from a board’s actual reflow profile. This is part of package classification and shipping labels linked to moisture exposure and reflow handling. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Translate classification profile as a required universal board recipe. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Consider the measurement envelope. A number belongs to a quantity, unit, sampling population and method. Removing one of those may create an apparent fact that was never measured, even if every digit survives intact. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Explain classification versus assembly process responsibility. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Check product datasheet and approved assembly profile. Inspect one ordinary case and one boundary case. The boundary case is often more revealing because it exposes a missing ‘only’, an unearned approval or an omitted exception that fluent target-language readers would not otherwise notice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 100 — Rebake decision

Fictional source situation. A lot is under review for permitted recovery handling after an exposure exception. This is part of package classification and shipping labels linked to moisture exposure and reflow handling. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. State that baking always restores acceptance. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Use an adversarial counterexample: imagine a reader who is motivated to release material, bypass a review or claim compliance. If the translation can reasonably be read as authorizing that behavior while the source cannot, it needs revision. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Retain pending authorized disposition and avoid operational advice. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Check supplier guidance and local procedure. Treat unresolved ambiguity as an explicit query to the owner. Translators can explain uncertainty, but they cannot create an engineering disposition, a supplier guarantee or a safety clearance through language choice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

21. Reliability qualification, stress and failure analysis

This domain concerns test evidence with stress conditions, sample populations and limited conclusions. Its characteristic translation risk is that passing one stress test is promoted into an unconditional reliability claim. The correct technical decision owner is reliability engineering and product quality, not the language editor. Before a translator selects a preferred equivalent, identify the controlled object, the source revision, the permitted human-readable display surface and any machine-readable value that must remain immutable. A single bilingual terminology note should include an unacceptable near-synonym, because knowing what a term must not mean is often more useful than memorizing its dictionary definition.

Diagnostic 101 — Qualification scope

Fictional source situation. A qualification report covers a defined package and process family. This is part of test evidence with stress conditions, sample populations and limited conclusions. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Say every future product variant is qualified. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Check the referent before checking the adjective. An identifier can refer to a wafer, a device, a carrier, a recipe or a document revision; choosing the wrong object means that even technically familiar vocabulary will describe the wrong evidence. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. State the supported configuration, sample and approved family boundary. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Compare qualification matrix and change history. The release reviewer should record which invariant was challenged, what source artifact resolved the doubt and which target-language line was approved. A ‘looks good’ response is not a suitable acceptance record for a controlled production context. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 102 — HTOL outcome

Fictional source situation. A high-temperature operating-life study meets its stated acceptance rule for a selected sample. This is part of test evidence with stress conditions, sample populations and limited conclusions. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Translate this as proof of infinite service life. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Read the sentence against its state transition. A request, observation, acknowledgment, execution and final approval are different events. A translation that jumps forward in that chain grants an action the source did not authorize. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Keep stress test, population and conclusion bounded. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Review test conditions, sample size and approved report. Keep the translated explanation outside executable keys or codes unless the actual schema authorizes a change. This separation lets human readers learn the meaning without making a localization update look like a manufacturing software revision. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 103 — HAST vs use

Fictional source situation. A humidity stress result is presented under accelerated test conditions. This is part of test evidence with stress conditions, sample populations and limited conclusions. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Present stress conditions as normal operating environment. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Consider the measurement envelope. A number belongs to a quantity, unit, sampling population and method. Removing one of those may create an apparent fact that was never measured, even if every digit survives intact. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Differentiate acceleration protocol from field usage. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Check report basis and user specifications. Inspect one ordinary case and one boundary case. The boundary case is often more revealing because it exposes a missing ‘only’, an unearned approval or an omitted exception that fluent target-language readers would not otherwise notice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 104 — Failure mode

Fictional source situation. Physical analysis identifies a damage signature but cause remains under investigation. This is part of test evidence with stress conditions, sample populations and limited conclusions. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Write that a specific supplier caused the failure. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Use an adversarial counterexample: imagine a reader who is motivated to release material, bypass a review or claim compliance. If the translation can reasonably be read as authorizing that behavior while the source cannot, it needs revision. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Preserve observation, hypothesis and verified root cause separately. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Reconcile microscopy evidence and approved investigation. Treat unresolved ambiguity as an explicit query to the owner. Translators can explain uncertainty, but they cannot create an engineering disposition, a supplier guarantee or a safety clearance through language choice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 105 — Screening vs qualification

Fictional source situation. A routine production screen is performed on every eligible unit; qualification used a sample plan. This is part of test evidence with stress conditions, sample populations and limited conclusions. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Exchange the terms because both involve tests. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Ask first what the source proves and what it deliberately leaves undecided. A translation should never fill that gap with optimism or with a harsher conclusion. This distinction matters when the next recipient must record a disposition rather than merely understand a paragraph. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Retain process purpose, population and authority. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Compare screening flow and qualification criteria. Save an independent comparison of source and target versions along with the test evidence. When the original changes, the record identifies which assumptions must be reopened rather than silently carrying old meanings forward. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

22. ESD handling, cleanroom and contamination controls

This domain concerns handling instructions that connect electrostatic and particulate risks to defined controls. Its characteristic translation risk is that a generic phrase may imply protection that the documented program does not claim. The correct technical decision owner is ESD coordinator, contamination control and EHS, not the language editor. Before a translator selects a preferred equivalent, identify the controlled object, the source revision, the permitted human-readable display surface and any machine-readable value that must remain immutable. A single bilingual terminology note should include an unacceptable near-synonym, because knowing what a term must not mean is often more useful than memorizing its dictionary definition.

Diagnostic 106 — ESD protected area

Fictional source situation. A sign identifies an area with defined ESD controls, not a space with zero discharge risk. This is part of handling instructions that connect electrostatic and particulate risks to defined controls. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Translate the label as electrostatic danger eliminated. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Read the sentence against its state transition. A request, observation, acknowledgment, execution and final approval are different events. A translation that jumps forward in that chain grants an action the source did not authorize. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Keep control-program scope and required behaviors. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Review IEC program and approved site training. Keep the translated explanation outside executable keys or codes unless the actual schema authorizes a change. This separation lets human readers learn the meaning without making a localization update look like a manufacturing software revision. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 107 — Gowning stage

Fictional source situation. A work instruction distinguishes gown entry sequence from a tool service procedure. This is part of handling instructions that connect electrostatic and particulate risks to defined controls. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Combine gowning and service authorization into one rule. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Consider the measurement envelope. A number belongs to a quantity, unit, sampling population and method. Removing one of those may create an apparent fact that was never measured, even if every digit survives intact. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Preserve personnel qualification and specific access prerequisite. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Check qualification record and access policy. Inspect one ordinary case and one boundary case. The boundary case is often more revealing because it exposes a missing ‘only’, an unearned approval or an omitted exception that fluent target-language readers would not otherwise notice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 108 — Cleanroom class

Fictional source situation. A report states a particular airborne-particle classification and measurement conditions. This is part of handling instructions that connect electrostatic and particulate risks to defined controls. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Describe the room as sterile. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Use an adversarial counterexample: imagine a reader who is motivated to release material, bypass a review or claim compliance. If the translation can reasonably be read as authorizing that behavior while the source cannot, it needs revision. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Keep particulate classification separate from microbiological sterility. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Check test standard and environmental report. Treat unresolved ambiguity as an explicit query to the owner. Translators can explain uncertainty, but they cannot create an engineering disposition, a supplier guarantee or a safety clearance through language choice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 109 — Particle trend

Fictional source situation. A trend plot indicates increasing counts without declaring a specification failure. This is part of handling instructions that connect electrostatic and particulate risks to defined controls. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Translate the increase as contamination event confirmed. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Ask first what the source proves and what it deliberately leaves undecided. A translation should never fill that gap with optimism or with a harsher conclusion. This distinction matters when the next recipient must record a disposition rather than merely understand a paragraph. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. State the observation and pending analysis. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Review measurement baseline and excursion rule. Save an independent comparison of source and target versions along with the test evidence. When the original changes, the record identifies which assumptions must be reopened rather than silently carrying old meanings forward. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 110 — ESD wrist strap

Fictional source situation. A monitoring screen reports a failed connectivity check on protective equipment. This is part of handling instructions that connect electrostatic and particulate risks to defined controls. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Translate it as routine operator information only. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Check the referent before checking the adjective. An identifier can refer to a wafer, a device, a carrier, a recipe or a document revision; choosing the wrong object means that even technically familiar vocabulary will describe the wrong evidence. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Preserve failed condition and appropriate stop/escalation signal. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Test warning recognition in a training UI. The release reviewer should record which invariant was challenged, what source artifact resolved the doubt and which target-language line was approved. A ‘looks good’ response is not a suitable acceptance record for a controlled production context. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

23. Equipment safety, hazardous materials and interlocks

This domain concerns safety documentation where response conditions and authority cannot be invented. Its characteristic translation risk is that unsafe imperatives may emerge from compression or localization. The correct technical decision owner is authorized equipment safety, EHS and emergency personnel, not the language editor. Before a translator selects a preferred equivalent, identify the controlled object, the source revision, the permitted human-readable display surface and any machine-readable value that must remain immutable. A single bilingual terminology note should include an unacceptable near-synonym, because knowing what a term must not mean is often more useful than memorizing its dictionary definition.

Diagnostic 111 — Interlock bypass

Fictional source situation. A manual explicitly forbids bypass without separately authorized controlled procedures. This is part of safety documentation where response conditions and authority cannot be invented. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Rewrite the exception as ordinary operator discretion. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Consider the measurement envelope. A number belongs to a quantity, unit, sampling population and method. Removing one of those may create an apparent fact that was never measured, even if every digit survives intact. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Keep prohibition and authority boundary without creating operational bypass steps. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Safety-owner approval and isolated comprehension test. Inspect one ordinary case and one boundary case. The boundary case is often more revealing because it exposes a missing ‘only’, an unearned approval or an omitted exception that fluent target-language readers would not otherwise notice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 112 — Exhaust status

Fictional source situation. An alarm indicates exhaust availability must be verified before a defined task. This is part of safety documentation where response conditions and authority cannot be invented. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Translate it as general ventilation advice. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Use an adversarial counterexample: imagine a reader who is motivated to release material, bypass a review or claim compliance. If the translation can reasonably be read as authorizing that behavior while the source cannot, it needs revision. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Preserve the blocking prerequisite and named responsibility. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Cross-check equipment hazard analysis and approved warning. Treat unresolved ambiguity as an explicit query to the owner. Translators can explain uncertainty, but they cannot create an engineering disposition, a supplier guarantee or a safety clearance through language choice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 113 — Chemical hazard

Fictional source situation. A supplier warning differentiates incompatible substances and emergency response contacts. This is part of safety documentation where response conditions and authority cannot be invented. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Combine hazards into vague ‘handle carefully’ text. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Ask first what the source proves and what it deliberately leaves undecided. A translation should never fill that gap with optimism or with a harsher conclusion. This distinction matters when the next recipient must record a disposition rather than merely understand a paragraph. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Keep hazard distinctions and route to trained EHS reviewers. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Compare controlled safety data and response procedure. Save an independent comparison of source and target versions along with the test evidence. When the original changes, the record identifies which assumptions must be reopened rather than silently carrying old meanings forward. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 114 — Lockout status

Fictional source situation. A maintenance ticket shows energy isolation requested but not verified. This is part of safety documentation where response conditions and authority cannot be invented. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Translate requested as isolation complete. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Check the referent before checking the adjective. An identifier can refer to a wafer, a device, a carrier, a recipe or a document revision; choosing the wrong object means that even technically familiar vocabulary will describe the wrong evidence. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Preserve physical verification as an independent required event. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Inspect signed isolation record, not just ticket status. The release reviewer should record which invariant was challenged, what source artifact resolved the doubt and which target-language line was approved. A ‘looks good’ response is not a suitable acceptance record for a controlled production context. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 115 — Emergency stop

Fictional source situation. A screen reports an emergency-stop function was activated and requires authorized assessment. This is part of safety documentation where response conditions and authority cannot be invented. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Translate it as a regular pause/resume prompt. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Read the sentence against its state transition. A request, observation, acknowledgment, execution and final approval are different events. A translation that jumps forward in that chain grants an action the source did not authorize. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Maintain the safety event and separate recovery authorization. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Run role-and-state comprehension tests without operating equipment. Keep the translated explanation outside executable keys or codes unless the actual schema authorizes a change. This separation lets human readers learn the meaning without making a localization update look like a manufacturing software revision. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

24. OT cybersecurity, remote service and audit trails

This domain concerns equipment computing evidence and security permissions across suppliers and operators. Its characteristic translation risk is that a multilingual interface can grant authority or erase incident history by wording alone. The correct technical decision owner is OT security, equipment owner and change control, not the language editor. Before a translator selects a preferred equivalent, identify the controlled object, the source revision, the permitted human-readable display surface and any machine-readable value that must remain immutable. A single bilingual terminology note should include an unacceptable near-synonym, because knowing what a term must not mean is often more useful than memorizing its dictionary definition.

Diagnostic 116 — Remote-access request

Fictional source situation. A vendor asked for temporary support access; review has not approved a session. This is part of equipment computing evidence and security permissions across suppliers and operators. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Describe the request as active remote control. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Use an adversarial counterexample: imagine a reader who is motivated to release material, bypass a review or claim compliance. If the translation can reasonably be read as authorizing that behavior while the source cannot, it needs revision. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Preserve pending approval, requested scope and session identity. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Check remote-access log and approver receipt. Treat unresolved ambiguity as an explicit query to the owner. Translators can explain uncertainty, but they cannot create an engineering disposition, a supplier guarantee or a safety clearance through language choice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 117 — Patch proposal

Fictional source situation. A proposed operating-system patch is under compatibility review. This is part of equipment computing evidence and security permissions across suppliers and operators. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Translate proposal as deployed security fix. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Ask first what the source proves and what it deliberately leaves undecided. A translation should never fill that gap with optimism or with a harsher conclusion. This distinction matters when the next recipient must record a disposition rather than merely understand a paragraph. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Keep test, approval and installation states distinct. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Compare software inventory and change record. Save an independent comparison of source and target versions along with the test evidence. When the original changes, the record identifies which assumptions must be reopened rather than silently carrying old meanings forward. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 118 — E187 revision

Fictional source situation. A 2026 standards article discusses a proposed tiered SEMI E187 revision. This is part of equipment computing evidence and security permissions across suppliers and operators. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. State that all proposed tiers are already mandatory. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Check the referent before checking the adjective. An identifier can refer to a wafer, a device, a carrier, a recipe or a document revision; choosing the wrong object means that even technically familiar vocabulary will describe the wrong evidence. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Distinguish current published specification from draft development. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Check dated SEMI source and current purchased edition. The release reviewer should record which invariant was challenged, what source artifact resolved the doubt and which target-language line was approved. A ‘looks good’ response is not a suitable acceptance record for a controlled production context. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 119 — Log retention

Fictional source situation. A multilingual admin label describes how to view an audit log, not permission to purge it. This is part of equipment computing evidence and security permissions across suppliers and operators. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Translate ‘view’ as ‘clear history’. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Read the sentence against its state transition. A request, observation, acknowledgment, execution and final approval are different events. A translation that jumps forward in that chain grants an action the source did not authorize. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Protect the action verb and record retention authority. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Test action permission matrix in a nonproduction interface. Keep the translated explanation outside executable keys or codes unless the actual schema authorizes a change. This separation lets human readers learn the meaning without making a localization update look like a manufacturing software revision. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 120 — End-of-life OS

Fictional source situation. The equipment inventory marks a component’s OS support as expired and a remediation plan pending. This is part of equipment computing evidence and security permissions across suppliers and operators. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Call it patched and secure because a plan exists. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Consider the measurement envelope. A number belongs to a quantity, unit, sampling population and method. Removing one of those may create an apparent fact that was never measured, even if every digit survives intact. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Retain current condition, unresolved exposure and planned mitigation. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Reconcile supported-version list and approved action plan. Inspect one ordinary case and one boundary case. The boundary case is often more revealing because it exposes a missing ‘only’, an unearned approval or an omitted exception that fluent target-language readers would not otherwise notice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

25. Customer notices, supplier changes and final disposition

This domain concerns technical customer communications with bounded claims and traceability obligations. Its characteristic translation risk is that polished wording may overpromise quality, compliance or delivered configuration. The correct technical decision owner is customer quality, supplier quality and product engineering, not the language editor. Before a translator selects a preferred equivalent, identify the controlled object, the source revision, the permitted human-readable display surface and any machine-readable value that must remain immutable. A single bilingual terminology note should include an unacceptable near-synonym, because knowing what a term must not mean is often more useful than memorizing its dictionary definition.

Diagnostic 121 — PCN effective date

Fictional source situation. A process-change notice announces a future effective point for affected parts. This is part of technical customer communications with bounded claims and traceability obligations. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Write that the change was already implemented across shipped goods. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Ask first what the source proves and what it deliberately leaves undecided. A translation should never fill that gap with optimism or with a harsher conclusion. This distinction matters when the next recipient must record a disposition rather than merely understand a paragraph. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Preserve applicability, notice date and first affected lot. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Check PCN identifiers and supply history. Save an independent comparison of source and target versions along with the test evidence. When the original changes, the record identifies which assumptions must be reopened rather than silently carrying old meanings forward. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 122 — Certificate scope

Fictional source situation. A certificate refers to one specified device lot and listed criteria. This is part of technical customer communications with bounded claims and traceability obligations. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Say the supplier certifies all product generations. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Check the referent before checking the adjective. An identifier can refer to a wafer, a device, a carrier, a recipe or a document revision; choosing the wrong object means that even technically familiar vocabulary will describe the wrong evidence. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Preserve exact product scope and criteria. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Compare certificate ID and shipment packing list. The release reviewer should record which invariant was challenged, what source artifact resolved the doubt and which target-language line was approved. A ‘looks good’ response is not a suitable acceptance record for a controlled production context. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 123 — Customer acceptance

Fictional source situation. A technical evaluation passed an internal review but customer approval is still pending. This is part of technical customer communications with bounded claims and traceability obligations. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Translate internal review as customer acceptance. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Read the sentence against its state transition. A request, observation, acknowledgment, execution and final approval are different events. A translation that jumps forward in that chain grants an action the source did not authorize. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Keep parties, decision authority and status distinct. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Check customer signoff and communication trail. Keep the translated explanation outside executable keys or codes unless the actual schema authorizes a change. This separation lets human readers learn the meaning without making a localization update look like a manufacturing software revision. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 124 — Containment notice

Fictional source situation. An investigation reports containment of potentially affected material; root cause remains open. This is part of technical customer communications with bounded claims and traceability obligations. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. State the problem is permanently solved. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Consider the measurement envelope. A number belongs to a quantity, unit, sampling population and method. Removing one of those may create an apparent fact that was never measured, even if every digit survives intact. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Describe containment, ongoing investigation and next report obligation. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Compare action plan and closure evidence. Inspect one ordinary case and one boundary case. The boundary case is often more revealing because it exposes a missing ‘only’, an unearned approval or an omitted exception that fluent target-language readers would not otherwise notice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 125 — Replacement authorization

Fictional source situation. A proposed replacement component awaits formal equivalence evaluation. This is part of technical customer communications with bounded claims and traceability obligations. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Translate similar function as approved interchangeable part. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Use an adversarial counterexample: imagine a reader who is motivated to release material, bypass a review or claim compliance. If the translation can reasonably be read as authorizing that behavior while the source cannot, it needs revision. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Keep equivalence conditional and exact part references. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Review qualification and contractual approval. Treat unresolved ambiguity as an explicit query to the owner. Translators can explain uncertainty, but they cannot create an engineering disposition, a supplier guarantee or a safety clearance through language choice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

26. Training interfaces, multilingual HMI and final handoff

This domain concerns human-readable content released alongside immutable manufacturing data. Its characteristic translation risk is that user interface comfort can conceal a changed command, warning or record. The correct technical decision owner is localization release manager, tool owner and training lead, not the language editor. Before a translator selects a preferred equivalent, identify the controlled object, the source revision, the permitted human-readable display surface and any machine-readable value that must remain immutable. A single bilingual terminology note should include an unacceptable near-synonym, because knowing what a term must not mean is often more useful than memorizing its dictionary definition.

Diagnostic 126 — Truncated button

Fictional source situation. A translated release-confirmation label is truncated and resembles an ordinary acknowledge button. This is part of human-readable content released alongside immutable manufacturing data. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Approve the layout because the full tooltip is correct. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Check the referent before checking the adjective. An identifier can refer to a wafer, a device, a carrier, a recipe or a document revision; choosing the wrong object means that even technically familiar vocabulary will describe the wrong evidence. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Keep the dangerous distinction visible without hover or extra context. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Test device width, role and keyboard navigation. The release reviewer should record which invariant was challenged, what source artifact resolved the doubt and which target-language line was approved. A ‘looks good’ response is not a suitable acceptance record for a controlled production context. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 127 — Colour-only state

Fictional source situation. A dashboard uses red and green icons alongside poorly differentiated translated labels. This is part of human-readable content released alongside immutable manufacturing data. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Assume operators can infer status from color alone. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Read the sentence against its state transition. A request, observation, acknowledgment, execution and final approval are different events. A translation that jumps forward in that chain grants an action the source did not authorize. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Ensure explicit text labels communicate held, running and released states. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Perform accessibility and grayscale comprehension tests. Keep the translated explanation outside executable keys or codes unless the actual schema authorizes a change. This separation lets human readers learn the meaning without making a localization update look like a manufacturing software revision. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 128 — Fallback language

Fictional source situation. A missing translation falls back to a source-language string in a safety panel. This is part of human-readable content released alongside immutable manufacturing data. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Treat the panel as fully localized. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Consider the measurement envelope. A number belongs to a quantity, unit, sampling population and method. Removing one of those may create an apparent fact that was never measured, even if every digit survives intact. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Flag incomplete critical content and provide authorized fallback behavior. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Audit all required locales and critical screen paths. Inspect one ordinary case and one boundary case. The boundary case is often more revealing because it exposes a missing ‘only’, an unearned approval or an omitted exception that fluent target-language readers would not otherwise notice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 129 — Training outcome

Fictional source situation. A trainee correctly explains a translated alarm but has not demonstrated tool qualification. This is part of human-readable content released alongside immutable manufacturing data. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Issue an operator authorization based on language comprehension alone. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Use an adversarial counterexample: imagine a reader who is motivated to release material, bypass a review or claim compliance. If the translation can reasonably be read as authorizing that behavior while the source cannot, it needs revision. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Keep localization training distinct from equipment certification. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Verify competency and authorization records separately. Treat unresolved ambiguity as an explicit query to the owner. Translators can explain uncertainty, but they cannot create an engineering disposition, a supplier guarantee or a safety clearance through language choice. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

Diagnostic 130 — Release receipt

Fictional source situation. Translation files passed syntax checks while a high-severity terminology query remains unresolved. This is part of human-readable content released alongside immutable manufacturing data. Identify exactly which record or physical object the statement covers. Before drafting a target-language sentence, mark the status, affected population, numerical evidence, time basis and named authority that appear in the original. Leave unsupported fields unknown rather than filling them with an engineer’s plausible assumption.

Attractive but defective interpretation. Declare the entire release approved. The fault is not a minor stylistic preference: it can transfer a property or decision from one object or stage to another. Ask first what the source proves and what it deliberately leaves undecided. A translation should never fill that gap with optimism or with a harsher conclusion. This distinction matters when the next recipient must record a disposition rather than merely understand a paragraph. An independent reviewer should explain why the defective interpretation is false before they read the preferred rendering; this prevents familiarity with the final answer from hiding the original ambiguity.

Meaning-preserving repair. Keep technical hold until the domain owner resolves the meaning query. Keep controlled identifiers, program references and numeric evidence exactly where their schemas require them. Rephrase human-readable connective language only to clarify the actual relationship. A good target sentence tells the reader what happened or may happen, what is not yet established and who has the authority to authorize any subsequent action. It need not be a literal word-for-word mirror, but it must not quietly change the production decision.

Verification and diagnostic. Inspect QA results, open issues, approved baseline and release signoff. Save an independent comparison of source and target versions along with the test evidence. When the original changes, the record identifies which assumptions must be reopened rather than silently carrying old meanings forward. For this case, write a two-column review note: one column names the source’s permissible conclusion; the other names an interpretation the source does not support. Reject the localization if those two columns become indistinguishable. This counterfactual check is especially valuable when a target-language word carries stronger permission or certainty than the original.

7. Transfer: from worked examples to independent translation capability

The purpose of the diagnostic collection is not to memorize a hundred and thirty particular answers. It is to develop a portable way of reading technical language. On every new job, ask what object the sentence refers to, which event or decision it describes, which evidence supports that description, which part is editable text and which part is an immutable data field. Once those questions are answered, the wording can be made clear for the target reader without producing a second, inconsistent manufacturing reality.

Transfer workshop 1: Reverse trace

Select a fictional released lot summary. Identify its wafer population, process revision, test-program revision and customer communication. Write four separate target-language sentences, one for each evidence source. Then reconstruct the full story and explain why no single sentence proves all four facts. Start by constructing a source-state ledger that lists the objects, applicable revision, time basis and available evidence. Before translating, state which conclusion the source actually supports and which tempting stronger conclusion it does not. This written distinction is the central skill being tested, not the ability to replace a word with a bilingual synonym.

Carry the exercise through a second review cycle. Ask one reviewer to test terminology and clarity; ask another to look specifically for changes in authorization, identity, numerical meaning and lifecycle state. Record the disagreement if the reviewers differ. The output is complete only when the target wording, source evidence, test case, review decision and unresolved questions form one traceable receipt. A learner who cannot produce that receipt should practice on fictional examples before handling live industrial documentation.

Transfer workshop 2: State-boundary drill

Build a decision table with requested, pending, observed, acknowledged, investigated, held, corrected and released. Write a neutral English gloss for each, then ask a domain reviewer which pairings are forbidden in the chosen workflow. Start by constructing a source-state ledger that lists the objects, applicable revision, time basis and available evidence. Before translating, state which conclusion the source actually supports and which tempting stronger conclusion it does not. This written distinction is the central skill being tested, not the ability to replace a word with a bilingual synonym.

Transfer workshop 3: Unit-integrity drill

Take a mock report with a temperature, time interval, layer thickness, concentration and electrical measurement. Preserve every quantity and unit. If locale formatting changes, show proof that the underlying stored number did not. Start by constructing a source-state ledger that lists the objects, applicable revision, time basis and available evidence. Before translating, state which conclusion the source actually supports and which tempting stronger conclusion it does not. This written distinction is the central skill being tested, not the ability to replace a word with a bilingual synonym.

Transfer workshop 4: Map-direction drill

Use a fictional 3-by-3 substrate grid with an off-center reference cell and two distinctive bins. Rotate and mirror hypothetical display renderings. Identify exactly which variants misassign device identities and document the approved source reference frame. Start by constructing a source-state ledger that lists the objects, applicable revision, time basis and available evidence. Before translating, state which conclusion the source actually supports and which tempting stronger conclusion it does not. This written distinction is the central skill being tested, not the ability to replace a word with a bilingual synonym.

Transfer workshop 5: Two-program test

Imagine two revisions of a wafer test program used on successive lots. Create a bilingual summary that keeps program revision and retest history separate. Explain why a combined yield figure needs its own denominator and approved method. Start by constructing a source-state ledger that lists the objects, applicable revision, time basis and available evidence. Before translating, state which conclusion the source actually supports and which tempting stronger conclusion it does not. This written distinction is the central skill being tested, not the ability to replace a word with a bilingual synonym.

Transfer workshop 6: Change-notice drill

Compare a proposed engineering change, an approved change with a future effective date and a completed implementation. Draft separate notices and ask reviewers to identify the exact first affected unit in each case. Start by constructing a source-state ledger that lists the objects, applicable revision, time basis and available evidence. Before translating, state which conclusion the source actually supports and which tempting stronger conclusion it does not. This written distinction is the central skill being tested, not the ability to replace a word with a bilingual synonym.

Transfer workshop 7: Safety-language drill

Rewrite a fictional warning three times for brevity, then verify that the prohibited act, prerequisite, severity and escalation contact remain visible. Never use training examples to design actual hazardous-equipment responses. Start by constructing a source-state ledger that lists the objects, applicable revision, time basis and available evidence. Before translating, state which conclusion the source actually supports and which tempting stronger conclusion it does not. This written distinction is the central skill being tested, not the ability to replace a word with a bilingual synonym.

Transfer workshop 8: Evidence-status drill

Take a supplier certificate, an internal test result and a customer acknowledgment. Make three target-language summaries. Explain why one cannot be casually translated as either of the others. Start by constructing a source-state ledger that lists the objects, applicable revision, time basis and available evidence. Before translating, state which conclusion the source actually supports and which tempting stronger conclusion it does not. This written distinction is the central skill being tested, not the ability to replace a word with a bilingual synonym.

Transfer workshop 9: HMI truncation drill

Put a critical state label in a narrow mobile or equipment-panel mock-up. Test its abbreviated form with people unfamiliar with the source language; reject any abbreviation that can mean both acknowledge and authorize. Start by constructing a source-state ledger that lists the objects, applicable revision, time basis and available evidence. Before translating, state which conclusion the source actually supports and which tempting stronger conclusion it does not. This written distinction is the central skill being tested, not the ability to replace a word with a bilingual synonym.

Transfer workshop 10: Security-control drill

Translate a remote-service request, an approved limited session and a completed access revocation. Keep permission, event time and operator role distinct; do not use a vulnerability memo as permission to bypass equipment controls. Start by constructing a source-state ledger that lists the objects, applicable revision, time basis and available evidence. Before translating, state which conclusion the source actually supports and which tempting stronger conclusion it does not. This written distinction is the central skill being tested, not the ability to replace a word with a bilingual synonym.

Transfer workshop 11: Glossary stress test

For hold, bin, lot, recipe, pass, release, qualification, monitor, outlier and deviation, write at least two distinct meanings that can occur in technical contexts. Make an approved sense selection depend on artifact and actor. Start by constructing a source-state ledger that lists the objects, applicable revision, time basis and available evidence. Before translating, state which conclusion the source actually supports and which tempting stronger conclusion it does not. This written distinction is the central skill being tested, not the ability to replace a word with a bilingual synonym.

Transfer workshop 12: Independent reviewer drill

Give a domain reviewer the translated text and the evidence only, not the original translation notes. Ask what action is permitted. A successful rendering makes the same bounded decision possible without the author’s explanation. Start by constructing a source-state ledger that lists the objects, applicable revision, time basis and available evidence. Before translating, state which conclusion the source actually supports and which tempting stronger conclusion it does not. This written distinction is the central skill being tested, not the ability to replace a word with a bilingual synonym.

8. How to build a terminology register without creating a shadow standard

Fab: Fabrication facility; not necessarily equivalent to a specific site or production line. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

Lot: Controlled grouping of material; can differ from wafer, carrier and equipment batch. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

Wafer: Substrate carrying many devices or patterns; one wafer is not one finished product. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

Die: Individual chip area on a wafer or separated device; identity must survive mapping. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

FOUP: Front-opening unified pod used for wafer transport; distinguish carrier and contents. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

Traveler: Record of authorized route and actual history; not merely an explanatory itinerary. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

Recipe: Controlled set of processing instructions and parameters; protect revision and ID. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

Route: Sequence or workflow; not automatically the physical path of a carrier. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

Mask: Patterning artifact; do not confuse mask identifier with localized legend. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

Overlay: Relative alignment measurement or representation; identify reference frame. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

Endpoint: Detected process endpoint; not automatically a product-release decision. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

Overetch: Named process segment; retain source definition and approval. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

Deposition: Addition of material or film; distinguish process target and measured outcome. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

Etch: Removal process; do not infer accepted part quality from completion. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

Implant dose: Implanted species quantity measure; not the same as implantation energy. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

Thermal budget: Cumulative thermal-exposure concept dependent on process context. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

Anneal: Thermal treatment; distinguish proposed step and completed event. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

CMP: Chemical mechanical polishing; process and specific quality measurements are separate. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

Metrology: Measurement science and measurement activities; not all inspection results are acceptance decisions. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

SPC: Statistical process control; control signals are not automatically specification failures. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

Control limit: Statistical process-monitoring boundary, with a defined basis. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

Specification limit: Product or process requirement boundary according to a controlling source. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

Excursion: Departure requiring evaluation under defined rule; scope and disposition must be stated. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

Hold: Controlled restriction pending disposition; not automatically reject or scrap. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

Disposition: Authorized determination of material treatment or status. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

MES: Manufacturing execution system; distinguish display labels from canonical records. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

SECS-II: Equipment message-content standard defined by SEMI E5. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

GEM: Equipment communication and control model defined by SEMI E30. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

HSMS: High-speed SECS message services; a transport context, not a general translation synonym. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

Collection event: Equipment-reporting event in the specific interface context. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

Alarm acknowledgment: Record of receipt or acknowledgment; not proof that a fault is corrected. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

Remote mode: Specific machine control mode; not automatic authorization to run. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

Carrier: Physical transport container or support; distinguish from wafer genealogy. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

Slot map: Association of physical carrier slots and material identities. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

Soft bin: Logical test classification whose mapping is program-dependent. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

Hard bin: Physical sort category or destination according to particular tester/handler mapping. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

Wafer map: Device-position information; orientation and origin are part of meaning. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

E142 substrate map: Structured device-location map whose schema and conventions must be checked. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

Retest: A second test event under defined rules, not automatic erasure of first result. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

Yield: Ratio requiring numerator, denominator, sample scope and method. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

Guardband: Conservative internal threshold separate from a published product limit. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

Qualification: Evidence process for defined scope; not a universal quality guarantee. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

Reliability: Time-related performance under conditions; not synonymous with test pass. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

HTOL: High-temperature operating life test; stress-test finding is bounded. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

HAST: Highly accelerated stress test; conditions are not ordinary end-use conditions. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

MSL: Moisture sensitivity level related to reflow handling and packaging. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

Dry pack: Protective packaging state; does not itself certify finished-device performance. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

ESD: Electrostatic discharge; control system has a defined scope and limits. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

Interlock: Protective control function; do not convert warning into permission to bypass. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

EHS: Environment, health and safety; review ownership is not ordinary proofreading. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

PCN: Product or process change notification, with applicability and effective dates. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

ECN: Engineering change notice; proposed, approved and effective statuses may differ. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

CoC: Certificate of conformity or compliance according to the actual contract and field. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

MTBF: Mean time between failures, with defined population and counting rule. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

MTTR: Mean time to repair, defined differently from time between failures. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

Audit trail: Chronological evidence of action, actor and time; do not silently rewrite. In a controlled project the register also needs the source quotation, approved target-language sense, forbidden near-synonym, object or system owner, effective revision and example of a failed interpretation. The purpose is to preserve a decision boundary, not to accumulate isolated translations.

9. Questions that distinguish useful translation from dangerous fluency

1. Can a machine translate a wafer-fab procedure automatically?

It can assist with first-pass language and terminology suggestions, but an authorized technical reviewer must validate instructions, units, revisions, safety language, states and machine-facing fields before controlled use. The operational check is to locate the corresponding source evidence and ask what the translated text would enable a reader to conclude or do. If the inference is stronger, broader or differently authorized, revise the wording and rerun review.

2. Is a SECS/GEM message name ordinary text?

Not when it functions as a protocol identifier or schema value. Translate human explanations in approved surfaces while preserving the code and the interface contract. The operational check is to locate the corresponding source evidence and ask what the translated text would enable a reader to conclude or do. If the inference is stronger, broader or differently authorized, revise the wording and rerun review.

3. Why is wafer-map translation risky?

The position, orientation, coordinate origin and association between die and bin are part of the evidence. A label or display change that relocates those associations is a data error. The operational check is to locate the corresponding source evidence and ask what the translated text would enable a reader to conclude or do. If the inference is stronger, broader or differently authorized, revise the wording and rerun review.

4. Does a passing wafer probe result mean a device can ship?

Not automatically. Product qualification, downstream assembly, final test, customer criteria and formal disposition may impose separate requirements. The operational check is to locate the corresponding source evidence and ask what the translated text would enable a reader to conclude or do. If the inference is stronger, broader or differently authorized, revise the wording and rerun review.

5. What should happen when a source term is ambiguous?

Record the ambiguity, cite its source location and ask the proper owner. Do not create a new operational meaning merely to make a fluent sentence. The operational check is to locate the corresponding source evidence and ask what the translated text would enable a reader to conclude or do. If the inference is stronger, broader or differently authorized, revise the wording and rerun review.

6. Can translators change measurement units?

Only when the approved deliverable permits it and the value, precision, basis and source record remain traceable. A conversion is not a casual typography correction. The operational check is to locate the corresponding source evidence and ask what the translated text would enable a reader to conclude or do. If the inference is stronger, broader or differently authorized, revise the wording and rerun review.

7. Do alarm acknowledgment and alarm clearance mean the same thing?

No. Receipt of an alarm can be recorded while the cause, equipment condition and corrective task remain unresolved. The operational check is to locate the corresponding source evidence and ask what the translated text would enable a reader to conclude or do. If the inference is stronger, broader or differently authorized, revise the wording and rerun review.

8. Why separate control limits from specification limits?

Control limits describe statistically established monitoring behavior; specification limits express requirements. They support different decisions and may use different evidence. The operational check is to locate the corresponding source evidence and ask what the translated text would enable a reader to conclude or do. If the inference is stronger, broader or differently authorized, revise the wording and rerun review.

9. What is the value of a forbidden-synonym glossary?

It makes a common, plausible but wrong meaning visible to reviewers, especially for words such as hold, pass, release, route and qualification. The operational check is to locate the corresponding source evidence and ask what the translated text would enable a reader to conclude or do. If the inference is stronger, broader or differently authorized, revise the wording and rerun review.

10. How should vendor acronyms be handled?

Use the supplier’s controlling interface or document glossary. Preserve the canonical key and ask when a proposed expansion is uncertain. The operational check is to locate the corresponding source evidence and ask what the translated text would enable a reader to conclude or do. If the inference is stronger, broader or differently authorized, revise the wording and rerun review.

11. Can a translated safety warning be shortened for a screen?

Only if an approved review demonstrates that hazard, prohibition, prerequisite and escalation remain clear. Brevity must not erase a safety condition. The operational check is to locate the corresponding source evidence and ask what the translated text would enable a reader to conclude or do. If the inference is stronger, broader or differently authorized, revise the wording and rerun review.

12. Does MSL imply long-term reliability?

No. It is a moisture/reflow sensitivity classification within a defined assembly-handling context and cannot be promoted into a lifetime guarantee. The operational check is to locate the corresponding source evidence and ask what the translated text would enable a reader to conclude or do. If the inference is stronger, broader or differently authorized, revise the wording and rerun review.

13. What if a standards article describes an upcoming revision?

Use the publication date and approval status. A proposed, balloted or discussed change should not be described as already mandatory. The operational check is to locate the corresponding source evidence and ask what the translated text would enable a reader to conclude or do. If the inference is stronger, broader or differently authorized, revise the wording and rerun review.

14. How do we verify a multilingual production screen?

Test representative tasks, user roles, narrow layouts, missing translations, state distinctions and boundary conditions in an authorized safe environment. The operational check is to locate the corresponding source evidence and ask what the translated text would enable a reader to conclude or do. If the inference is stronger, broader or differently authorized, revise the wording and rerun review.

15. What should a translation release receipt contain?

Source and target revisions, language, owner, approved terminology version, unresolved-query outcome, test results, reviewers and final authorization. The operational check is to locate the corresponding source evidence and ask what the translated text would enable a reader to conclude or do. If the inference is stronger, broader or differently authorized, revise the wording and rerun review.

16. How are fictional examples different from real instructions?

Fictional cases train interpretation but are not plant-approved procedures. Their values, wording and sequences must not be used to operate real manufacturing equipment. The operational check is to locate the corresponding source evidence and ask what the translated text would enable a reader to conclude or do. If the inference is stronger, broader or differently authorized, revise the wording and rerun review.

17. Do suppliers and fabs use exactly the same glossary?

Not necessarily. A standard provides common terms, while equipment specifications, plant workflows and contracts define implementation details. The operational check is to locate the corresponding source evidence and ask what the translated text would enable a reader to conclude or do. If the inference is stronger, broader or differently authorized, revise the wording and rerun review.

18. Should failed tests disappear after an approved retest?

No. The approved decision may rely on retest rules, but the history of both observations and their authorization remains relevant. The operational check is to locate the corresponding source evidence and ask what the translated text would enable a reader to conclude or do. If the inference is stronger, broader or differently authorized, revise the wording and rerun review.

19. Can we translate a manufacturing part number?

Not when it is an immutable identifier. A descriptive product label may have a translation, but the part number remains canonical. The operational check is to locate the corresponding source evidence and ask what the translated text would enable a reader to conclude or do. If the inference is stronger, broader or differently authorized, revise the wording and rerun review.

20. What makes an industrial translation genuinely accurate?

A target reader can identify the same physical objects, evidence, conditions, constraints, uncertainties and authorized next actions as the source reader. The operational check is to locate the corresponding source evidence and ask what the translated text would enable a reader to conclude or do. If the inference is stronger, broader or differently authorized, revise the wording and rerun review.

10. Primary standards and authoritative terminology routes

The sources below establish scope and vocabulary for readers; they are not substitutes for licensed standard text, product-specific procedures or professional approval. The cited versions and revision statuses were checked against public sources on 30 October 2026. If a supplier’s approved interface contract and a general description differ in applicability, escalate the difference rather than treating the general webpage as an operating instruction.

SEMI Information and Control standards index: Identifies SEMI E5 (SECS-II), E30 (GEM), E37 (HSMS), E87 carrier management, E90 substrate tracking and related equipment standards.
SEMI introduction to SECS/GEM: Explains distinction between machine communication model, message content and transport.
SEMI E142 substrate mapping specification: Public product overview identifies the 0225 edition and the subject of substrate mapping.
SEMI E142 revision activity, April 2026: Describes proposed coordinates transformation work; a proposal is not automatically an implemented requirement.
SEMI E10 and E79 performance webinar: Distinguishes RAM, utilization, OEE and throughput measurement.
SEMI safety guidelines index: Identifies S1 equipment labels, S2 equipment EHS, S6 exhaust and other safety guidelines.
IEC 61340-5-1:2024: Current ESD control requirements for handling susceptible electronic equipment; scope and exceptions matter.
IPC/JEDEC J-STD-020F:2022: Moisture/reflow sensitivity classification, distinct from general reliability assurance.
SEMI E187 published security specification: Describes the published 0122 baseline for equipment cybersecurity.
SEMI E187 proposed 2026 changes: Explains work toward tiered security levels; do not misstate a pending revision as already binding.

11. Next routes: advance without overwriting the parent systems

Begin with the original Master Art of Translation | The Complete System for Moving Meaning Between Languages when the problem involves broad translation practice, equivalence, revision control or the relationship between source and target meaning. Return here when the question is specifically about semiconductor production records, wafer-test meaning, equipment-interface fields, substrate mapping and downstream chip traceability.

For deliberate practice in semantic distinctions, consult the protected Vocabulary Learning Hub; it owns general vocabulary learning rather than fab workflows. For interpretation, sentence relations and language across contexts, follow How English Works | Translation and Meaning Across Languages. Those resources are linked, not retitled or replicated.

The practical next step is to select one genuine but non-sensitive source artifact, build its immutable-token list and object graph, run the sixteen gates, then invite a domain expert to attempt to disprove the translated decision. Document what the expert finds. Translation becomes trustworthy not when the words sound impressive, but when the evidence, constraints and authorized actions survive being read in another language.