Construction Drawings & BIM Translation | Master Art of Translation

How do you translate construction drawings, architectural plans, building specifications and BIM models without changing what must be built? Construction drawing translation is not simply replacing labels on a plan. A reliable translation must preserve the dimensions, units, grid references, revision status, material requirements, drawing legends, element identities and approval boundaries that enable designers, contractors, fabricators and facility managers to read the same technical intention.

This complete construction and BIM translation guide explains how to translate architectural drawings, structural plans, mechanical and electrical schedules, quantity take-offs, specifications, requests for information, building information modelling (BIM) properties, Industry Foundation Classes (IFC) data and operation-and-maintenance records. It answers practical questions such as how to translate a construction specification, how to localize BIM object names without changing identifiers, how to prevent errors in drawing notes, how to maintain revisions across languages and how to quality-check multilingual construction documentation.

The repeatable method begins by finding the controlling source, identifying the decision each document supports, separating translatable prose from protected codes and geometry, mapping terminology to approved concepts, checking every numerical relationship, validating consistency across drawings and models, and releasing only traceable target-language versions. The examples below are fictional teaching cases. This is a world-facing framework, not a substitute for applicable building law, professional design, engineering approval, construction supervision or any project-specific contractual hierarchy.

Reading route: This specialised application belongs to Master Art of Translation | The Complete System for Moving Meaning Between Languages. It concentrates on engineering information and building delivery rather than repeating the general language foundation in the Vocabulary Learning Hub or the How English Works learning map. The narrower companion construction drawings, specifications and site-instructions translation guide remains a separate owner; this page extends the method into linked digital models, information management, handover and multi-document verification.

Safety boundary: no example below is an approved construction detail or regulatory-compliant design. Never derive a buildable detail or substitute for the design professional’s judgement from a translated sentence alone. A mismatch involving structural capacity, fire and life safety, electrical isolation, accessibility, equipment commissioning or a field instruction is a reason to stop, preserve the evidence and request a competent decision.

1. Search intent: what readers are actually trying to solve

A person searching for “translate construction drawings” can be a bilingual engineer, an architect exchanging files across a language boundary, a project manager seeking multilingual coordination, a translator deciding whether to touch a technical abbreviation, or a building owner receiving handover documents from a geographically dispersed supply chain. Their apparent request is linguistic. Their practical request is operational: which facts must remain stable so that someone can interpret, purchase, install, inspect, maintain or reject the correct thing? Those are different questions, and the second governs the first.

Search phrases such as architectural plan translation, building drawing localization, BIM terminology translation, technical specification translation, construction document translation services, IFC property translation, multilingual building information modelling, translating MEP drawings, translating bill of quantities, and how to check construction translation reveal distinct jobs. A tender translation helps compare proposals. A fabrication note helps make a component. An as-built record helps locate an installed component later. A maintenance instruction enables a future operator to isolate a system and find its asset tag. Treating these as a single generic text job hides the decisions the words carry.

The result a reader needs is not a collection of translated nouns. It is a reliable chain from source information to target interpretation, with the source version, named responsibilities, measured values, approval state, model properties and field verification intact. In high-consequence contexts a translation may be fluent yet wrong because it strengthens a recommendation into a requirement, changes a provisional issue into a construction release, converts a clear dimension into a nominal dimension or joins two differently defined quantities.

This guide is organised as a mechanism and workshop. Read the first sections to understand why technical meaning behaves differently from ordinary prose. Use the gates to organise work. Study the document-family playbooks to locate the risk. Use the worked diagnostics as rehearsal exercises, not ready-made bilingual construction instructions. End with validation templates, transfer tests and routes back to the translation system.

2. The hidden problem: a drawing is a network of decisions, not a picture with words

A construction drawing is a structured assertion about a proposed or existing physical object. Lines, dimensions, line types, symbols, labels, sections, references and notes have interacting roles. The same word may name a geometric element, a specification section, a material property, a procurement item or an operational asset; it cannot be translated correctly until the role is known. A translator who sees only extracted text may not know whether “fall” concerns a floor gradient, a fall-prevention risk, a roof slope or a fall in pressure.

A plan also has authority, but not every plan has the same authority. A sketch issued for discussion, a coordinated model, an approved shop drawing, a tender addendum and a record drawing can present similar geometry while carrying different instructions. Publication dates, model status, drawing revisions and transmittal purposes matter because they tell the reader which interpretation may be relied upon. Translating “for review” as “approved” is not a small wording error; it converts a request for a judgement into a fictitious judgement.

There is a second hidden problem: information moves through different representations. An architect’s schedule may say “D-14”; an IFC exchange may associate an element with a global identifier; a contractor’s procurement list may identify a manufacturer model; an operations system may give the installed door a separate asset tag. If translation overwrites these identifiers with readable but invented labels, the building loses traceability. Conversely, refusing to translate human-facing descriptions can exclude workers who need to interpret what the protected code refers to.

The correct target is therefore a meaning-preserving map: maintain immutable identifiers and geometry, translate descriptions that genuinely support comprehension, preserve reference links and qualified conditions, and expose unresolved conflicts instead of quietly reconciling them. This is as much an information-governance problem as a language problem. The most expensive translation failure may be one that looks polished enough to pass an ordinary reading review.

3. The mechanism: six layers that must remain connected

The first layer is identity: a drawing number, revision, object GUID, grid marker, equipment tag or specification identifier. Identity answers “which record or component?” It often must be copied exactly, not rendered into a conversational synonym. A bilingual glossary can explain a tag, but must not change the tag as though it were ordinary text.

The second layer is geometry and quantity: measured length, level, angle, count, capacity, tolerance, flow rate, thickness, quantity basis or associated unit. A figure without its measurement basis is not a complete figure. Values must be compared as structured facts, not visually skimmed for similar-looking digits. The reader must know whether a dimension is nominal, clear, centre-to-centre, finished, minimum, maximum or indicative.

The third layer is classification and properties: a door’s function, a building element’s material, a system’s discipline, a property set or a classification code. Model names are not always reliable property definitions. ISO 12006-2:2015 supplies a classification framework but does not itself supply every operational classification table. ISO 23386:2020 addresses methodology for properties in interconnected dictionaries. A project must still decide which actual dictionaries and identifiers control.

The fourth layer is decision authority: who issues information, who approves it, which revision supersedes another, what is proposed rather than decided, and what remains subject to confirmation. The fifth is evidence: the record supporting an assertion, from a submittal and test certificate to a survey result or commissioning record. The sixth is lifecycle purpose: whether the information is used for planning, design, fabrication, construction, inspection, handover or maintenance.

Every translation should preserve all six layers that are present. A clear natural-language sentence is successful only if the same authorised actor can make the same supported decision from it. If an element’s name changes while its GUID stays stable, that may be a legitimate localized display. If the GUID is replaced, or if the condition on a quantity disappears, the conversion has broken the information system rather than translated it.

4. International references: standards as anchors, not automatic approval

The references are intentionally high-authority and purpose-limited. ISO 19650-1:2018 explains concepts and principles for information management using building information modelling across an asset lifecycle. ISO 19650-2:2018 concerns information management in the delivery phase; ISO’s public catalogue notes that a replacement is under development, which is different from claiming that a draft has replaced the published edition. These references are useful for organising responsibilities, exchange states and versions. They do not say that any particular translated drawing is acceptable for construction.

ISO 16739-1:2024 describes the IFC open data schema for exchanging building and infrastructure information, including entities, property and quantity sets and exchange structures. A text localization must respect the difference between display names, coded values, entity types, property definitions and stable identifiers. IFC is not a promise that two applications will preserve every meaning automatically; export settings and supported views still need examination.

ISO 23386:2020 gives a methodology to define, author and maintain properties in interconnected data dictionaries. ISO 12006-2:2015 provides a framework for classification and was confirmed in 2026. ISO 128-1:2020 outlines general principles for technical drawing representation and was also confirmed in 2026. ISO 21542:2021 addresses accessibility and usability of the built environment; its review status as of 2026 must not be confused with an amended design requirement for a particular project. ISO 19650-6:2025 covers collaborative management of structured health and safety information, but this article does not replace its contents with invented rules.

The buildingSMART Information Delivery Specification (IDS) v1.0 was endorsed as a standard in 2024. IDS can help express and check information requirements in an IFC model; it does not by itself judge whether an engineering design complies with law. These references are not copied technical standards, and this guide avoids pretending to reproduce text that requires licensed access. Its purpose is to show where terminology, identifiers and validation logic should be checked.

5. A repeatable fifteen-gate method for translation that stays buildable only when authorised

Gate 1 — Define the task. Establish whether the deliverable is an informative bilingual view, a contractual translation, an annotated review, a machine-readable model export, a tender interpretation or an approved operational document. Do not let a translation silently become an approval, an instruction or a design revision.

Gate 2 — Freeze the source. Record the controlling file, sheet, model, revision, issue purpose, date, discipline and document hierarchy. A screenshot without status is insufficient where the same drawing has been issued several times.

Gate 3 — Identify the authoritative owner. Determine the designer, responsible engineer, model information manager, specification author or field authority capable of resolving disputed meaning. Translators should not choose a structural interpretation because it sounds natural.

Gate 4 — Build an identity ledger. Extract sheet identifiers, equipment tags, material marks, grid references, IFC GUIDs, property keys, test report numbers and links to detail sheets. Mark them as stable identifiers unless the controlling system explicitly authorises otherwise.

Gate 5 — Extract numbers as facts. Capture value, unit, precision, condition, comparison sign, tolerance, reference surface and measurement basis. Copying “1:100” accurately but confusing drawing scale with a slope ratio would fail this gate.

Gate 6 — Parse the grammar of obligation. Distinguish shall, must, should, may, proposed, subject to approval, not less than, not exceeding, unless and only after. Keep the actor and the condition attached to the right action.

Gate 7 — Establish terminology ownership. Ask whether a term is a general word, a technical concept defined by a standard, a project-specific definition or a contract-specific capitalised expression. Choose an approved concept and record non-equivalent alternatives.

Gate 8 — Read the visual neighbours. For a note on a section, inspect the referenced view, section cut, line legend, schedule and corresponding model object. Optical text extraction alone can hide which dimension or component a leader line actually identifies.

Gate 9 — Translate by semantic unit. Translate clauses, tables, callouts, symbols and property descriptions with their referents visible. Do not translate a detached string containing “typical” without knowing which detail it qualifies.

Gate 10 — Protect structure. Keep numbering, links, cross-references, object keys, data types, units, classification codes, table relations and revision labels traceable. Localize human-facing labels separately from machine-readable identifiers.

Gate 11 — Compare independent evidence. Test against the related schedule, specification, design report, issue register, bill of quantities, model and transmittal. When sources disagree, label the discrepancy and request a decision; never fabricate alignment.

Gate 12 — Review high-risk categories. Prioritise structural assumptions, fire protection, egress, accessibility, safe isolation, pressure and flow specifications, temporary works, lifting and any interpretation where an action cannot easily be reversed.

Gate 13 — Test target-language users. A qualified reviewer should answer questions about the affected component, condition, value, authority and next action without consulting the source language. Distinguish linguistic readability from design approval.

Gate 14 — Release with provenance. Package source hash or equivalent identifier, version, translation revision, approval scope, glossary version, resolved issues, outstanding queries, language and reviewer. Never strip those fields for cosmetic neatness.

Gate 15 — Retest after changes. A new drawing issue, IFC export, supplier substitution or site clarification can invalidate only some translated elements, but a targeted delta review must find them. Old translated files should not remain visually indistinguishable from the new authoritative set.

These gates form a practical control loop: identify, preserve, verify, authorise, release and reassess. They are a translation quality system, not a new construction code. The governing decision remains with the responsible project authorities.

6. The failure atlas: twenty ways meaningful-looking words corrupt an information set

Failure 1: an issue-purpose adjective disappears, making an information-only file look issued for construction. Failure 2: a preliminary dimension loses its qualifier. Failure 3: a minimum clearance becomes an approximate dimension. Failure 4: a datum is mistaken for a floor level. Failure 5: an element tag is translated as though it were an ordinary label.

Failure 6: a quantity converts without the underlying measurement basis. Failure 7: a tolerance becomes a nominal target. Failure 8: a specification clause’s exception is detached from its obligation. Failure 9: a “typical unless noted” condition becomes a universal requirement. Failure 10: an abbreviation is expanded using the wrong discipline’s vocabulary.

Failure 11: a model property identifier is replaced when only its display name should change. Failure 12: a revision description is taken as approval evidence. Failure 13: an inspection result is rewritten as design verification. Failure 14: a proposed product is represented as accepted. Failure 15: a site observation is mislabelled as an instruction to change work.

Failure 16: a fire-resistance or acoustic rating is equated across regimes without checking the relevant classification basis. Failure 17: a schematic route is treated as a surveyed location. Failure 18: a required field is silently omitted from an IFC export. Failure 19: a digital asset tag is lost between practical completion and operations. Failure 20: the target-language document preserves every number but attaches one of them to the wrong object.

These errors are not solved by making the prose longer. Detect them by checking referents, action conditions, source authority, measurement basis, identifiers and cross-document evidence. A reviewer should be able to say not merely “this word is inaccurate” but “this change would cause the reader to act on the wrong component, status, requirement or evidence state”.

7. Field playbooks: twenty-four linked document families

The following twenty-four playbooks are not interchangeable. Each gives a different object of translation, different evidence and different failure modes. They demonstrate how to reason across architectural, structural, MEP, BIM and lifecycle records while leaving engineering authority with the competent project team.

1. Document control, issue purpose and revision authority

Document control is the first physical-risk boundary because identical geometry may circulate for briefing, tender, coordination, fabrication, construction and record purposes. Translate the status as a claim about the permitted use of information; never as a judgement on the quality of the drawing. The key crosscheck is the title block against the transmittal and the common data environment.

2. Views, sections, dimensions and annotation references

The visual grammar of drawings answers where a note applies. A callout, line type, section arrow and dimension extension line must remain attached to the same feature after text is localized. An elegant bilingual annotation set that moves the target of an instruction is less trustworthy than a less polished version with perfect referencing.

3. Grids, datums, coordinates and levels

A coordinate is useful only with its reference framework. Local grid names, global coordinates, survey control, floor labels and reduced levels may coexist, and translation must not merge them into a single everyday notion of height or position. Preserve coordinate reference systems and the original labels until the competent survey or design owner resolves equivalence.

4. Structural assumptions, load cases and performance limits

Technical drawings often summarize assumptions that are developed in design reports. A translation cannot recalculate or certify these assumptions. Its obligation is to keep load category, combination, direction, unit and stated limits legible to the engineer. Distinguishing serviceability from ultimate strength is important; changing one to another would change the purpose of a check.

5. Concrete, reinforcement and cast-in provisions

Concrete information combines geometry, material specification, sequencing and inspection requirements. A translator must separate grade designations, bar marks, spacing, cover, construction joints, pour breaks and embedded items. The most important question is which physical or procedural constraint a phrase defines, not which literal noun looks closest in a dictionary.

6. Steelwork, fabrication marks and connection records

Fabrication notes carry shop identifiers, member profiles, bolt assemblies, weld symbols and inspection classifications. A project may use different standards in different contracts, so a translator should not invent a supposedly universal steel designation. Preserving the actual member mark and test evidence is often more important than choosing the most familiar term.

7. Timber, composites and material conditioning

Natural and composite construction materials carry moisture, grade, orientation, exposure and manufacturer-specific information. A word such as ‘dry’ may refer to a treatment or acceptance condition, not an aesthetic description. Translation needs to keep the stated evidence and avoid choosing an unverified substitute specification from a different material system.

8. Door, window and opening schedules

Schedules join element IDs, room relations, dimensions, performance claims, hardware and interface details. The words ‘door size’ can hide a distinction among rough opening, nominal leaf, clear passage and overall frame dimensions. Translation must carry the exact column definition and keep each element’s identity in the correct row.

9. Fire compartments, smoke control and egress records

Life-safety translation must resist the temptation to make complex ratings sound simpler. Fire resistance, reaction to fire, smoke leakage, means of escape and emergency wayfinding are not the same property. Each rating belongs to a test, classification, system or jurisdiction. Language clarification must never manufacture a blanket compliance claim.

10. Accessibility, reach, circulation and usability

Accessibility claims require particular care because a superficially similar term may refer to a different functional requirement. Dimensions, approach geometry, turning and reach conditions, contrast and route continuity need their own qualifiers. Refer to the controlling accessibility authority; the international reference helps locate concepts but cannot automatically approve local compliance.

11. Facade geometry, thermal performance and weather exposure

A facade is a system of layers, interfaces, supports and performance evidence. Terms like ‘sealed’, ‘insulated’ and ‘weatherproof’ are not universal passes. Translation should say which tested or calculated property applies, to what assembly, and under which exposure or installation conditions. Component-level data cannot automatically become building-level performance.

12. Roofing, waterproofing and drainage details

Roof and waterproofing documents combine slope, membrane continuity, outlets, junctions and maintenance access. A translation that preserves every word but drops an exception at a penetration can break the intended water path. Do not convert a concept drawing into an approved waterproofing detail. The correct diagnostic traces water and document references, not merely vocabulary.

13. Heating, ventilation and cooling systems

Mechanical diagrams use airflow direction, pressure, balancing quantities, equipment tags and control sequences. A note may be schematic rather than an exact route. Good translation keeps the diagram’s kind, not just its individual nouns, and resists translating a setpoint or airflow unit into an assumed comfort target.

14. Plumbing, sanitary and gravity drainage

Pipework records distinguish service, route, connection, flow direction, pipe dimensions, gradient, installation elevation and tested capacity. Similar-looking numerical values may relate to invert levels, nominal diameter, pressure class or flow, and those variables must never be conflated in the target language.

15. Electrical distribution, lighting and control wiring

Electrical translation needs strict separation between labels, functional states and actionable instructions. Circuit identifiers, phase designations, voltage, protection settings, connection diagrams and isolation records must be interpreted by competent electrical personnel. A language reviewer cannot verify safe operation from prose alone.

16. Fire detection, suppression and specialist systems

Systems integration uses alarm interfaces, zones, cause-and-effect logic and equipment certification. The translation should make conditions comprehensible without rewriting control logic. Distinguish design specification from test evidence and ensure the responsible systems engineer checks the multilingual document.

17. Geotechnical assumptions, foundations and below-grade work

Below-ground information is particularly vulnerable to false certainty. Borehole observations, interpreted ground models, assumed bearing parameters, design groundwater levels and foundation details have different evidence statuses. Translation must preserve whether a figure was measured, inferred, assumed or verified during excavation.

18. Surveys, earthworks and site geometry

Earthworks information may mix preconstruction surface survey, design surface, permitted excavation, actual formation and final as-built conditions. A translation that drops the surface source can turn a quantity estimate into a claim about measured excavation. Maintain coordinate reference and sequence before translating elevation labels.

19. Quantities, measurement rules and commercial schedules

Measurement tables are not merely lists of things. A quantity may be net, gross, provisional, measured, scheduled, paid or excluded by the governing measurement rules. Translation must preserve the measurement boundary and should never choose a contractual pricing interpretation without the commercial document owner.

20. Submittals, product data and material substitutions

A submittal package links a proposed product to an approved specification, required test evidence and an acceptance decision. A translation must separate manufacturer’s claims from the designer’s review and the contractor’s procurement choice. Trade names, model identifiers, configuration and certification scope should remain explicit.

21. Site instructions, RFIs, variations and responsibility boundaries

The same sentence can be a question, a recommendation, a contractual variation request or a binding instruction depending on its issuer and workflow status. A translator should document that speech act and keep any unresolved value visible. This avoids converting a multilingual discussion into an unapproved site change.

22. IFC data, stable object identity and semantic dictionaries

BIM translation can damage a model while leaving its geometry visually intact. IFC property names, entity classes, GUIDs, type instances and material classifications are structured data, not ordinary marketing labels. A localized user interface can present friendly names while leaving machine-referenced identifiers unchanged. ISO 16739-1 and ISO 23386 are especially relevant to this boundary.

23. Common data environment, information delivery and interoperability

Controlled information exchange relies on a record of what was submitted, what is shared, what was accepted for a defined purpose and what remains in progress. ISO 19650 guides information-management thinking, while project-specific CDE conventions define actual statuses and field names. The translator must not invent correspondence between unfamiliar platform labels.

24. Commissioning, asset handover and facilities management

At handover, the purpose of information changes from design intention to operational evidence. Installed asset tag, serial number, test result, warranty condition, replacement part and maintenance interval often live in different systems. Translation should preserve those joins. A polished bilingual manual is poor if an operator cannot find the exact installed machine.

8. Worked diagnostics: ninety-six fictional source-to-decision repairs

Each exercise is an invented fragment, not a specimen of a legally binding drawing or a recommended installation detail. The source wording is deliberately compact so the reviewer can see the exact place where meaning might be changed. Work through the stated problem, original meaning, repair and verification evidence before moving to the next case. Some cases need a language correction; others need a qualified design response because the source cannot support a safe interpretation.

Domain 1: Document control, issue purpose and revision authority

Diagnostic 001: Coordination is not construction

Source situation. Sheet A-204 Rev C marked ‘issued for coordination’. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of document control, issue purpose and revision authority, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. Rev C is translated as approved for construction. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: the statement identifies a coordination-stage use, not a construction permission. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Retain the issue-purpose phrase and mark its authority separately from the revision letter. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Compare sheet title block, issue register and transmittal before releasing bilingual copies. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic transfer. Imagine this phrase copied from a drawing into a contractor’s meeting minutes. Ask whether a participant could incorrectly assume permission to proceed, an altered size or a closed issue. If so, the localization must keep the limiting qualifier in the same sentence or provide an explicit cross-reference. Record the decision before the next document package is issued.

Diagnostic 002: Superseded detail

Source situation. An old detail S-311 Rev A remains attached to a later package. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of document control, issue purpose and revision authority, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translator calls it the latest detail because its file creation date is recent. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: creation and issue dates are not a replacement for the controlling revision chain. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Keep the sheet ID and original revision; flag supersession rather than silently editing the detail. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Check the document register, design change log and package contents. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic transfer. Now move the same information from a PDF into an IFC property or maintenance table. Ask which part is a machine key, which is display text, and which part expresses a condition. A successful bilingual export preserves the original connection; a translation that only preserves the visible wording fails this test.

Diagnostic 003: Review comment mistaken for direction

Source situation. A review note reads ‘consider shifting door location subject to designer confirmation’. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of document control, issue purpose and revision authority, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The target text says move the door. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: the note asks for evaluation and does not authorise a relocation. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Keep ‘consider’ and ‘subject to confirmation’ and attribute the decision to the designer. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Follow the comment, response and closed-out design issue. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic transfer. Consider a reviewer comparing two revisions in different languages. The user should still find the same source item, understand which version controls and identify what evidence would settle an ambiguity. If an apparent translation correction actually changes design intent, it belongs in a documented engineering revision instead.

Diagnostic 004: Drawing stamp meaning

Source situation. A submittal has a ‘reviewed with comments’ stamp and a dated contractor copy. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of document control, issue purpose and revision authority, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translation changes the stamp to fully accepted. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: a review classification does not automatically remove unresolved comments or contractual boundaries. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Translate the exact stamp and preserve the comment attachment reference. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Inspect the actual stamp legend and related submittal decision. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic transfer. Ask a target-language reader to state the affected object, the relevant value or state, the allowed inference and the next responsible reviewer. Do not prompt them with the expected answer. A confident but unsupported response exposes a communication defect even when the translated vocabulary is technically sophisticated.

Domain 2: Views, sections, dimensions and annotation references

Diagnostic 005: Plan view versus elevation

Source situation. A note refers to a section at grid B/4, not an elevation drawn beside it. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of views, sections, dimensions and annotation references, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The note is placed on the adjacent view in the target-language file. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: the reader would associate the correct words with the wrong geometric feature. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Keep the section reference and callout placement, and treat view labels as linked identifiers. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Review the sheet graphically with both views visible. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 006: Dimension line ownership

Source situation. A 900 mm clear opening is dimensioned between finished faces. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of views, sections, dimensions and annotation references, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translation calls 900 mm the door leaf width. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: clear opening and leaf width measure different things. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. State the finished-face measurement basis without introducing a manufacturing dimension. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Compare door schedule, wall finish detail and dimension extension lines. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 007: Scale versus slope

Source situation. A detail title says scale 1:20 while a drainage arrow indicates a gradient. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of views, sections, dimensions and annotation references, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The ratio 1:20 is narrated as the floor slope. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: the scale describes drawing representation rather than physical drainage geometry. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Attach each ratio to its symbol and location; do not infer a gradient from the drawing scale. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Check sheet legend and dimensioned gradient note. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 008: Typical detail exception

Source situation. A callout reads ‘TYP. except at movement joint’. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of views, sections, dimensions and annotation references, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translated note omits the exception. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: typical treatment would wrongly extend into a location explicitly excluded. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Keep the exception next to the callout and identify the specific movement-joint detail. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Trace related details and the drawing note index. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Domain 3: Grids, datums, coordinates and levels

Diagnostic 009: Structural grid identity

Source situation. Column C7 stands at grid intersection C/7. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of grids, datums, coordinates and levels, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The column tag C7 is rewritten as a translated word-plus-number label. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: grid references function as stable technical keys rather than prose. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Preserve C/7 and C7 exactly while translating the explanatory sentence. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Compare plan, schedule, setting-out model and survey reference. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 010: Reference level

Source situation. The plan states +12.350 relative to the project datum. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of grids, datums, coordinates and levels, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The target text calls it 12.350 metres above sea level. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: project and geodetic datums are not interchangeable. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Retain the stated datum basis and ask for a conversion only if a surveyed transformation exists. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Check datum declaration, benchmark register and level annotations. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 011: Slab top and finished floor

Source situation. A section labels structural top of slab and a separate finished floor elevation. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of grids, datums, coordinates and levels, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translator uses one generic floor-height label for both. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: the finish build-up separates structural and occupied surfaces. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Retain each surface name and its distinct elevation value. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Compare architectural finish schedule and structural section. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 012: Surveyed versus indicative boundary

Source situation. A plan shows an approximate site line pending survey confirmation. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of grids, datums, coordinates and levels, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translation states a precise property boundary. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: indicative graphics are not verified cadastral evidence. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Retain approximate status and the need for a licensed survey confirmation. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Check survey report, legend and applicable legal record. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Domain 4: Structural assumptions, load cases and performance limits

Diagnostic 013: Serviceability limit

Source situation. A note refers to a serviceability deflection criterion. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of structural assumptions, load cases and performance limits, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The target text changes it to a failure-load criterion. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: serviceability and ultimate capacity address different performance questions. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Use the approved terminology and retain the same criterion symbol and context. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Cross-check structural design note and calculation reference. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 014: Line load and area load

Source situation. A schedule distinguishes kN/m from kN/m². This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of structural assumptions, load cases and performance limits, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translation removes the denominator in both units. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: the quantities have different physical dimensions and cannot be substituted. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Preserve unit strings and define what element or surface each quantity applies to. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Compare load schedule, legend and analysis assumptions. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 015: Temporary load assumption

Source situation. A construction-stage diagram shows a temporary load case pending engineer review. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of structural assumptions, load cases and performance limits, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The target version presents it as a permanent design allowance. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: stage and approval status determine the meaning of the assumption. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Keep the temporary-stage qualification and request the engineer’s decision. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Inspect sequence diagram, design report and reviewed calculation. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 016: Directional force

Source situation. A bracing note specifies resistance in the indicated axis. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of structural assumptions, load cases and performance limits, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The target sentence simply says resist lateral force. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: the removed axis may change which member or connection is intended. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Preserve directional reference and any orientation arrow. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Verify model local axes, note leader and related connection detail. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Domain 5: Concrete, reinforcement and cast-in provisions

Diagnostic 017: Bar spacing

Source situation. A reinforcement note uses 150 mm c/c. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of concrete, reinforcement and cast-in provisions, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translation says 150 mm clear gap. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: centre-to-centre spacing is not the space between bar surfaces. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Preserve the centre-to-centre measurement basis and bar mark. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Crosscheck reinforcement schedule and detail notation. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 018: Concrete cover

Source situation. A detail notes nominal cover measured to specified reinforcement. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of concrete, reinforcement and cast-in provisions, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translation describes total member thickness. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: cover and section depth are separate dimensions. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Retain the covered surface and the reinforcement element being measured. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Compare section view, bar layout and cover schedule. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 019: Construction joint

Source situation. A pour sequence labels a construction joint distinct from an expansion joint. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of concrete, reinforcement and cast-in provisions, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. Both become the same generic joint word. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: the two joint functions have different design and execution consequences. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Adopt project-approved joint terminology and keep detail cross-references. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Check pouring sequence and joint specification. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 020: Cast-in anchor

Source situation. A schedule lists an embedded anchor assembly requiring placement before the pour. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of concrete, reinforcement and cast-in provisions, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The note is rendered as an after-fix drilling instruction. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: the timing and anchorage system have been changed. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Preserve embedded status and the pre-pour coordination requirement. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Check approved shop drawings, hold points and supplier detail. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Domain 6: Steelwork, fabrication marks and connection records

Diagnostic 021: Fabrication mark

Source situation. Beam B-27 is tagged with a shop assembly identifier. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of steelwork, fabrication marks and connection records, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. B-27 is translated as beam number twenty-seven without its mark context. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: the mark links cutting lists, inspection and erection sequences. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Retain B-27 as a literal key and translate only the descriptive heading. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Compare erection drawing, shop drawing and delivery label. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 022: Weld symbol

Source situation. An arrow-side weld symbol is associated with a joint detail. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of steelwork, fabrication marks and connection records, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translation describes a weld on the opposite side. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: the annotation’s geometry is part of its meaning. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Leave the symbol intact and preserve the interpretation with the qualified fabrication reviewer. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Inspect symbol standard, arrow position and detail. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 023: Bolt grade

Source situation. The note lists a specified bolt property class and coating. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of steelwork, fabrication marks and connection records, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translation replaces it with a generic high-strength bolt. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: property class and surface treatment are not interchangeable shorthand. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Keep the stated classification and coating details precisely. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Check fastener schedule and material certificate. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 024: Inspection status

Source situation. A weld report records inspection completed with a defect pending disposition. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of steelwork, fabrication marks and connection records, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The target summary says inspection passed. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: completion of inspection does not imply acceptance of findings. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Retain defect status and disposition requirement without changing the factual test result. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Check inspection report, NCR and final disposition. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Domain 7: Timber, composites and material conditioning

Diagnostic 025: Timber moisture

Source situation. A product sheet gives a moisture-content test range. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of timber, composites and material conditioning, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translation says the timber is completely dry. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: measured moisture content is not an absolute absence of water. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Preserve range, units and testing context. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Compare supplier certificate and acceptance clause. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 026: Grain direction

Source situation. A panel detail specifies orientation relative to a supporting span. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of timber, composites and material conditioning, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translated instruction drops the grain direction. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: the material’s structural behaviour may depend on orientation. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Keep orientation terms and their diagram references. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Check panel manufacturer detail and installation plan. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 027: Composite rating

Source situation. A facade product has a test report for a named assembly. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of timber, composites and material conditioning, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translation attributes its rating to every product of that brand. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: test evidence applies to a specific tested system and conditions. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Retain assembly identification and limits of the test scope. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Compare configuration, report number and declared product. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 028: Surface treatment

Source situation. A specification distinguishes factory-applied protection from site-applied finish. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of timber, composites and material conditioning, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translation treats them as one coating. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: sequence, warranty and responsibility may differ. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Use separate terms for factory treatment and field finishing. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Compare finish schedule, submittal and maintenance information. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Domain 8: Door, window and opening schedules

Diagnostic 029: Door clear width

Source situation. The accessibility note states minimum clear passage width. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of door, window and opening schedules, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The target-language schedule translates this as nominal door leaf width. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: passage width depends on geometry and hardware conditions. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Keep the defined clear-width concept and specified minimum. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Review opening schedule and accessibility detail. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 030: Fire door assembly

Source situation. A door set has a test report covering leaf, frame and hardware combination. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of door, window and opening schedules, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translation describes only the door leaf as tested. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: assembly performance must not be simplified into a component claim. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Retain assembly scope and configuration information. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Crosscheck test report, hardware schedule and approved product. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 031: Glazing notation

Source situation. A window schedule distinguishes glazing specification from frame finish. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of door, window and opening schedules, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The target text merges both into a single material description. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: different components have different specification and replacement paths. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Keep separate glazing and frame fields with the original window ID. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Check product schedule and detail keynotes. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 032: Door swing

Source situation. A plan uses a swing arc and handedness notation. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of door, window and opening schedules, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translated note reverses left and right without defining viewing side. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: handedness depends on the reference convention and cannot be guessed. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Keep the diagram and original hand code; ask for a defined target convention. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Verify plan, door schedule and hardware supplier definition. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Domain 9: Fire compartments, smoke control and egress records

Diagnostic 033: Compartment boundary

Source situation. A line on a fire strategy plan identifies a compartment boundary. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of fire compartments, smoke control and egress records, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The target text calls it a decorative partition. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: the boundary communicates a safety function not visible from surface appearance. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Retain the compartment classification and drawing legend. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Compare fire strategy document and architecture detail. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 034: Rating shorthand

Source situation. A schedule contains EI 60 under a specified classification regime. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of fire compartments, smoke control and egress records, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The target version equates it to an unrelated rating code. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: equivalence requires evidence and may not exist across test regimes. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Preserve the original rating and identify the governing standard. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Check classification report and responsible fire designer. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 035: Smoke-control sequence

Source situation. A control diagram links a damper state to a defined alarm condition. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of fire compartments, smoke control and egress records, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translation omits the triggering condition. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: a control response without its trigger changes functional behaviour. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Preserve signal, state and conditional sequence without editing logic. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Compare cause-and-effect matrix and commissioning sequence. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 036: Escape route designation

Source situation. A diagram distinguishes a designated escape route from a general corridor. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of fire compartments, smoke control and egress records, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translation labels both as ordinary access. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: a route’s life-safety purpose is lost. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Keep the controlled designation and referenced wayfinding symbol. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Crosscheck life-safety plan and signage schedule. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Domain 10: Accessibility, reach, circulation and usability

Diagnostic 037: Clear turning space

Source situation. A floor plan reserves an unobstructed turning space. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of accessibility, reach, circulation and usability, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translation says the indicated area is optional circulation. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: the reserved space is a defined functional requirement. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Keep unobstructed status and the geometric envelope. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Check accessible-route plan and coordinated furniture layout. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 038: Reach height

Source situation. A schedule records maximum operable control reach height. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of accessibility, reach, circulation and usability, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The target-language value is described as a recommended average. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: a maximum restriction has become a nonbinding average. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Preserve comparison sign and user interaction context. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Cross-check equipment mounting and applicable design criteria. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 039: Ramp gradient

Source situation. A ramp note differentiates slope from crossfall. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of accessibility, reach, circulation and usability, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translation swaps the two terms. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: travel gradient and cross-gradient control different aspects of access. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Keep arrow direction and corresponding ratio paired. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Review section, plan and accessibility report. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 040: Threshold

Source situation. A door detail specifies threshold profile subject to a defined accessible transition. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of accessibility, reach, circulation and usability, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translation describes any raised sill as acceptable. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: the profile has functional and dimensional constraints. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Retain exact threshold detail number without deriving a substitute. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Check approved detail and accessibility review record. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Domain 11: Facade geometry, thermal performance and weather exposure

Diagnostic 041: U-value context

Source situation. A facade schedule identifies thermal transmittance for a specified assembly. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of facade geometry, thermal performance and weather exposure, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translation attributes the number to glass alone. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: assembly and component thermal metrics may be different. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Retain test or calculation boundary and units. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Compare energy report and exact facade system. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 042: Movement allowance

Source situation. A curtain-wall detail specifies accommodation of structural movement. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of facade geometry, thermal performance and weather exposure, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translation turns it into a fixed gap requirement. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: movement behaviour cannot be reduced to a single nominal opening. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Keep movement direction and performance condition. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Compare structural movement criteria and facade detail. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 043: Sealant compatibility

Source situation. A note requires sealant compatibility confirmation with adjacent materials. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of facade geometry, thermal performance and weather exposure, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translation states the named sealant is approved for all substrates. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: a pending compatibility review is a conditional decision. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Preserve approval condition and named interfaces. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Check supplier documentation and material submittal. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 044: Air-water tests

Source situation. A report distinguishes air leakage and water penetration results. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of facade geometry, thermal performance and weather exposure, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The target summary uses one generic leakage term. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: different tests measure distinct performance mechanisms. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Keep the test condition, quantity and outcome for each phenomenon. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Review test method and laboratory report. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Domain 12: Roofing, waterproofing and drainage details

Diagnostic 045: Roof falls

Source situation. A roofing plan marks arrows toward named outlets. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of roofing, waterproofing and drainage details, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translation associates an arrow with a downpipe material note. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: direction and destination are geometric facts. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Preserve outlet identity and arrow relationship. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Compare roof plan, drainage schedule and section. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 046: Membrane laps

Source situation. A detail specifies lap orientation relative to exposure. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of roofing, waterproofing and drainage details, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The target phrase says overlap generally without orientation. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: the orientation is a distinct installation condition. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Retain arrow and assembly-specific orientation terminology. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Check approved waterproofing supplier system. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 047: Penetration detail

Source situation. A service pipe crossing the roof references a named flashing detail. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of roofing, waterproofing and drainage details, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translation replaces the detail reference with generic sealing language. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: a detailed interface has been replaced by an unapproved technique. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Keep the detail number and request clarification if source is missing. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Review penetration schedule and coordination model. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 048: Overflow route

Source situation. A roof drainage note distinguishes normal outlet from emergency overflow path. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of roofing, waterproofing and drainage details, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. Both become ‘drain’ without roles. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: the secondary path may be needed under a different operating condition. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Retain both roles and their exact outlets. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Verify drainage calculations, plans and schedules. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Domain 13: Heating, ventilation and cooling systems

Diagnostic 049: Supply versus extract

Source situation. An air distribution schedule separates supply and extract terminal airflows. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of heating, ventilation and cooling systems, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. Both are rendered as ventilation intake. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: flow direction is part of the system function. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Keep supply and extract terms and equipment tag associations. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Check schematic and air balance schedule. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 050: Setpoint and alarm threshold

Source situation. A controls sequence has a temperature setpoint and a separate high alarm threshold. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of heating, ventilation and cooling systems, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translation replaces both with target temperature. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: commanded operating value and alarm condition serve different purposes. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Preserve each variable name, units and triggering comparison. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Compare control points list and approved sequence. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 051: Schematic routing

Source situation. A duct diagram is explicitly marked schematic. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of heating, ventilation and cooling systems, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translation represents the drawn alignment as an exact installation route. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: schematic intent does not establish surveyed spatial position. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Retain schematic qualifier and coordination requirement. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Check coordinated model and construction drawings. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 052: Filter classification

Source situation. An equipment schedule lists a filter class according to a named method. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of heating, ventilation and cooling systems, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The target version substitutes a familiar marketing label. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: different classifications are not automatically identical. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Keep the classification system and exact designation. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Check manufacturer’s test data and project schedule. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Domain 14: Plumbing, sanitary and gravity drainage

Diagnostic 053: Nominal pipe size

Source situation. A plumbing schedule gives DN 100. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of plumbing, sanitary and gravity drainage, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translation says measured internal bore is exactly 100 mm. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: nominal designation is not necessarily an actual measured internal diameter. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Retain DN designation and specification basis. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Check pipe standard, schedule and product datasheet. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 054: Invert level

Source situation. A drainage profile labels invert level at a manhole. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of plumbing, sanitary and gravity drainage, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translation calls it the cover level. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: top-of-cover and pipe invert are distinct surveyed elevations. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Retain the level type and reference datum. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Check profile, chamber schedule and survey control. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 055: Gravity flow

Source situation. A drain plan has a flow direction arrow and slope notation. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of plumbing, sanitary and gravity drainage, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The target note reverses direction to match reading order. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: the physical route cannot be changed by sentence layout. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Keep arrow, start and end identifiers coupled. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Review profile and longitudinal section. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 056: Pressure-test record

Source situation. A test report records pressure maintained during a defined test period. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of plumbing, sanitary and gravity drainage, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translation says the system is permanently leak-free. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: a bounded test outcome does not prove every future operating condition. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Preserve test duration, method, result and acceptance decision. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Compare signed test sheet and commissioning register. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Domain 15: Electrical distribution, lighting and control wiring

Diagnostic 057: Circuit tag

Source situation. Panel DB-2 circuit 14 is named in a cable schedule. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of electrical distribution, lighting and control wiring, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translator renames it distribution board fourteen. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: panel ID and outgoing circuit number are different keys. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Keep DB-2 and circuit 14 as separate identifiers. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Crosscheck single-line diagram and cable register. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 058: Normally open

Source situation. A contact is labelled normally open in the defined de-energised state. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of electrical distribution, lighting and control wiring, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translated text says it must always remain open. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: contact state terminology is not an operating command. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Preserve the defined reference condition and schematic symbol. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Review control diagram and device datasheet. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 059: Emergency lighting

Source situation. A schedule distinguishes normal lighting from emergency provision. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of electrical distribution, lighting and control wiring, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. Both are labelled general illumination. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: operational purpose and testing pathway differ. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Retain functional category and supply source. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Compare lighting plan and emergency systems schedule. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 060: Protection setting

Source situation. A report lists a protective device setting pending coordination approval. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of electrical distribution, lighting and control wiring, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translation makes it a final commissioned setting. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: provisional calculation differs from authorised configuration. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Keep pending status, unit and device identifier. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Check approved coordination study and test record. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Domain 16: Fire detection, suppression and specialist systems

Diagnostic 061: Alarm zone

Source situation. A detector is assigned to zone Z-03 on a control matrix. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of fire detection, suppression and specialist systems, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translation gives it the room number instead. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: zone and room reference serve different information systems. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Preserve Z-03 and its descriptive location separately. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Check detection drawings and control panel zone list. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 062: Valve state

Source situation. A schematic states valve normally closed with a defined alarm operation. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of fire detection, suppression and specialist systems, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translation instructs operators to keep it closed during all alarms. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: normal state and event response are different conditions. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Retain exact state logic and avoid inventing an operating order. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Compare cause-and-effect matrix and vendor sequence. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 063: Suppression coverage

Source situation. A design narrative defines a protected area for a named suppression system. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of fire detection, suppression and specialist systems, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The target sentence claims whole-building coverage. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: scope has been broadened beyond the design. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Keep protected-area boundaries and exceptions. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Check fire strategy drawings and system calculations. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 064: Test completion

Source situation. A commissioning record says a functional test was witnessed with an open issue. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of fire detection, suppression and specialist systems, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The summary says the system has full approval. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: witnessing, acceptance and close-out may occur separately. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Preserve issue status and named reviewer. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Check punch list and signed commissioning certificate. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Domain 17: Geotechnical assumptions, foundations and below-grade work

Diagnostic 065: Borehole observation

Source situation. A log records groundwater encountered during drilling on one date. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of geotechnical assumptions, foundations and below-grade work, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The target document calls it permanent maximum groundwater level. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: one observation does not establish an extreme design condition. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Preserve date, method and observation wording. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Check investigation report and groundwater monitoring data. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 066: Ground model assumption

Source situation. A foundation note says strata depth is interpreted between boreholes. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of geotechnical assumptions, foundations and below-grade work, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translation treats every interpolated line as an observed contact. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: inferred geometry has been presented as direct measurement. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Retain interpreted status and drawing legend. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Check investigation cross-sections and design assumptions. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 067: Pile cut-off

Source situation. A piling schedule distinguishes design toe level and pile cut-off level. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of geotechnical assumptions, foundations and below-grade work, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The two are merged as pile depth. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: toe and cut-off describe different ends of an element. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Keep each level, datum and pile ID. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Check pile schedule, geotechnical drawing and installation record. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 068: Bearing verification

Source situation. An inspection note records a foundation bearing surface awaiting engineer acceptance. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of geotechnical assumptions, foundations and below-grade work, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The target phrase says bearing has been certified. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: inspection evidence is not yet a completed acceptance decision. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Preserve open hold point and named authority. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Review inspection record and final release. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Domain 18: Surveys, earthworks and site geometry

Diagnostic 069: Cut-and-fill estimate

Source situation. A model summary reports estimated cut volume from a design surface. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of surveys, earthworks and site geometry, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translation describes it as measured excavated material. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: design-derived quantities and recorded production quantities differ. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Retain estimate status and surface comparison basis. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Check volume report, survey date and model version. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 070: Benchmark

Source situation. A setting-out plan references benchmark BM-04. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of surveys, earthworks and site geometry, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translator localizes BM-04 into a name resembling another marker. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: control point identifiers must remain exact. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Keep BM-04 and clarify its role in ordinary language. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Check survey control report and site register. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 071: Retaining interface

Source situation. A site section shows an excavation limit beside an existing asset. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of surveys, earthworks and site geometry, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The target-language note treats the limit as an allowable unsupported cut. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: geometry alone does not state a temporary works method. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Preserve the limit and any design or temporary-work qualification. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Refer to support design and approved sequence. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 072: As-built deviation

Source situation. A survey record reports position deviation from design. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of surveys, earthworks and site geometry, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translation erases the deviation and prints the intended location. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: recorded state and intended design are distinct evidence. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Preserve measured and design coordinates as separate fields. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Crosscheck survey record and acceptance disposition. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Domain 19: Quantities, measurement rules and commercial schedules

Diagnostic 073: Net versus gross area

Source situation. A bill of quantities distinguishes net finished floor area from gross built area. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of quantities, measurement rules and commercial schedules, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translation uses one term for both. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: areas use different inclusion and exclusion rules. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Keep calculation basis and item references. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Check measurement rules, schedule and drawings. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 074: Provisional quantity

Source situation. A tender schedule states quantity is provisional. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of quantities, measurement rules and commercial schedules, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The target text says the quantity is final. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: a provisional basis has been converted into a commitment. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Retain provisional status and adjustment terms. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Check tender conditions and quantity schedule. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 075: Rate and amount

Source situation. A cost table has quantity, rate, currency and total amount columns. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of quantities, measurement rules and commercial schedules, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translation labels rate as total cost. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: the numeric relationship may remain plausible while the column meaning changes. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Preserve each column function and currency. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Recalculate a few sample row relationships without adjusting the contract. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 076: Exclusion

Source situation. A measurement note excludes temporary supports from a particular item. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of quantities, measurement rules and commercial schedules, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translated note removes the exclusion clause. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: scope and pricing may be expanded. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Retain the exclusion and cross-reference its own item. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Check contractual measurement notes and linked items. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Domain 20: Submittals, product data and material substitutions

Diagnostic 077: Proposed substitution

Source situation. A contractor proposes alternative product M-18 subject to consultant approval. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of submittals, product data and material substitutions, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translation describes M-18 as approved equivalent. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: proposed equivalence is a claim requiring technical acceptance. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Retain proposed status and approving role. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Check submittal, technical comparison and decision. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 078: Certificate scope

Source situation. A product certificate applies to a specific material configuration. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of submittals, product data and material substitutions, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The target summary omits model and thickness restrictions. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: evidence may not cover the item being procured. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Keep certificate number and applicability limits. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Compare installed item, declared configuration and certification. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 079: Sample approval

Source situation. A finish sample was selected for colour only. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of submittals, product data and material substitutions, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The target document presents it as acceptance of the whole product system. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: aesthetic selection may not settle fire, strength or durability requirements. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Preserve limited approval purpose. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Check sample record and remaining product submittal requirements. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 080: Manufacturer instruction

Source situation. An installation guide describes a warranty condition under a specified substrate. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of submittals, product data and material substitutions, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translation generalises it to all substrates. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: the conditional performance claim was broadened. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Retain substrate and preparation condition in the explanatory translation. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Verify manufacturer’s current version and project-approved details. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Domain 21: Site instructions, RFIs, variations and responsibility boundaries

Diagnostic 081: RFI question

Source situation. An RFI asks whether a penetration location may shift. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of site instructions, rfis, variations and responsibility boundaries, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The target note says shift the penetration. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: a question seeking authorisation became an instruction. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Preserve interrogative form and the pending answer. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Check RFI response and authorised design change. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 082: Field observation

Source situation. A supervisor notes a discrepancy between site condition and drawing. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of site instructions, rfis, variations and responsibility boundaries, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translated report instructs remedial demolition. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: observation is evidence, not necessarily a selected remedy. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Retain observational status and requested technical review. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Check field report and issued instruction. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 083: Variation quotation

Source situation. A proposed variation lists time and cost impacts. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of site instructions, rfis, variations and responsibility boundaries, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translation calls it an agreed change order. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: commercial proposal and approved variation differ. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Keep proposal status and approval dependencies. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Check change register and signed authorisation. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 084: Instruction supersession

Source situation. A new site instruction explicitly supersedes an earlier clarification. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of site instructions, rfis, variations and responsibility boundaries, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translation simply adds both to the final handbook. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: conflicting directions cannot be presented as concurrently valid. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Mark the superseded instruction and reference the controlling issue. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Review instruction register, issue dates and acknowledged changes. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Domain 22: IFC data, stable object identity and semantic dictionaries

Diagnostic 085: IFC GUID

Source situation. An IFC wall object has a globally unique identifier. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of ifc data, stable object identity and semantic dictionaries, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translation process regenerates the identifier when changing its name. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: identity linking review, cost and handover records is broken. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Translate display text only; protect GUID and relationships. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Compare GUID counts and object links before and after export. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 086: Property name

Source situation. A model uses a controlled property key and a translatable descriptive caption. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of ifc data, stable object identity and semantic dictionaries, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. Both key and caption are renamed into target language. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: downstream validators may no longer recognise the property. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Preserve schema key and localize a separate human-facing label. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Compare property dictionary and exported IFC. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 087: Type versus occurrence

Source situation. A door type has a general classification and individual instances have unique marks. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of ifc data, stable object identity and semantic dictionaries, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The target dataset gives every instance the same tag. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: type-level properties and occurrence identity were collapsed. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Maintain type identifier alongside each occurrence identifier. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Check door schedule and element-instance counts. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 088: Null property

Source situation. A required property is absent in the source model. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of ifc data, stable object identity and semantic dictionaries, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translation inserts an assumed default value to appear complete. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: unknown information has been fabricated. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Leave null or missing state explicit and flag it as a data-quality issue. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Run IDS or project validation and request the model owner. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Domain 23: Common data environment, information delivery and interoperability

Diagnostic 089: CDE suitability status

Source situation. An information container has status ‘shared for coordination’. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of common data environment, information delivery and interoperability, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translator maps it to an approved construction status. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: suitability and release purpose have changed. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Maintain the project-specific status definition and original code. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Consult CDE status matrix and information-management protocol. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 090: Package manifest

Source situation. A transmittal references model, drawings and schedule from a single issue. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of common data environment, information delivery and interoperability, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The target package includes a newer model but the old sheet list. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: documents no longer represent one coherent exchange. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Preserve and crosscheck the manifest identifiers and revision matrix. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Reconcile export log, issue package and transmittal. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 091: IDS requirement

Source situation. An IDS check requires a named property for certain IFC elements. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of common data environment, information delivery and interoperability, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translation describes all missing properties as compliant. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: validation failure has been concealed. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Retain applicability, property identity and pass-fail evidence. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Run the same rule against the delivered model and compare reports. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 092: Open issue

Source situation. A model clash is logged as unresolved with responsible discipline named. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of common data environment, information delivery and interoperability, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translated issue register marks it resolved after a discussion. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: a discussion is not a closed and checked issue. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Preserve state, owner, location and evidence of closure. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Inspect issue history and federated model review. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Domain 24: Commissioning, asset handover and facilities management

Diagnostic 093: Design versus installed model

Source situation. The design schedule names a proposed pump while the as-built schedule records a substitute. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of commissioning, asset handover and facilities management, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translation uses the original design model for the operation manual. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: the maintainer may identify the wrong equipment. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Use as-built asset identity and preserve record of substitution. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Check signed as-built register and installed nameplate. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 094: Serial number

Source situation. A maintenance register lists a unique serial number and a separate product series. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of commissioning, asset handover and facilities management, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translator translates or normalizes the serial number. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: traceability to the physical asset is lost. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Copy the serial exactly and translate only the descriptive product family. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Compare nameplate photo, register and warranty record. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 095: Test result

Source situation. A commissioning sheet records a measured result within one test configuration. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of commissioning, asset handover and facilities management, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The translated manual states the device always achieves that performance. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: a bounded result has become an unconditional guarantee. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Keep condition, instrument, date, result and acceptance status. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Check original commissioning record and operating limits. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

Diagnostic 096: Preventive maintenance

Source situation. A maintenance schedule distinguishes an inspection interval from a replacement interval. This is an invented project record, not a complete drawing note. The translator’s first job is to identify what kind of record this is and what a reader is entitled to infer from it. The record belongs to the domain of commissioning, asset handover and facilities management, where a seemingly small change of definition can reach another schedule, model, report or physical element.

Tempting but incorrect rendering. The target text turns both into compulsory replacement. The mistake is plausible because a general-language reading can appear smoother or more familiar than the source. It is still unacceptable as a controlled technical translation: different maintenance actions, resources and risks are conflated. The reviewer should name precisely what is lost—a measurement basis, a conditional obligation, an identifier, a functional state, a source limitation or an approval boundary—rather than merely declaring the sentence awkward.

Meaning-preserving repair. Retain task, trigger, frequency and responsible role separately. Preserve the actual source notation wherever it functions as a key and translate only the words necessary for a reader to understand the controlled meaning. If the source and related project records conflict, enter a query with the original text, affected identifier, document status and proposed readings. The translator must not invent a technical decision to make the target-language version look complete.

Independent check. Compare OEM manual and approved asset maintenance plan. The quality reviewer should locate the exact source field, match its referent in the adjoining view or linked dataset, record the agreed terminology and confirm the target sentence supports the same action as the authoritative record. This is a comparison of evidence rather than a vote on which sentence sounds better.

9. Full dossier walkthrough: one multilingual design-to-handover chain

Consider a fictional project with an early architectural plan, a structural model, a door schedule, a fire strategy note, a procurement submittal, an inspection record and an asset handover file. The first plan shows a door as D-14, issued for coordination. The preliminary door schedule carries a nominal leaf size and a separate minimum clear opening. The fire strategy describes the intended performance of a specific door assembly. A supplier proposes a tested configuration, and the designer’s response is pending.

The translator first assembles the document manifest rather than translating the whole folder immediately. Each file receives a row showing owner, issue status, revision, source language, intended use and links to the others. The door tag D-14, associated IFC identity and specification clause references are protected. The schedule’s two kinds of width are recorded as two distinct data facts. If a conflicting drawing appears, the translator raises it rather than guessing which number the design team meant.

The next task is to classify statements by what they can prove. A drawing label may show intended location. A specification may state a performance requirement. A laboratory report may establish that one tested configuration met a named classification under a stated test. A submittal response may approve a specified product within a defined scope. An installation inspection may record what was actually installed. These are six different evidence states. Translation should preserve them rather than summarising all of them as “door compliant.”

Suppose the architectural plan then moves to a new revision, changing the adjacent wall finish. The original geometric model, the door clear opening and the hardware schedule may all need coordination, but the translator should not calculate a new compliant clear width. Instead, the change is traced to the relevant design owner. The target-language file gains a change log stating which translation fields are provisional, which source issue is current and which earlier bilingual package is no longer controlling.

At procurement, a supplier’s model number enters a distinct identity ledger. It must not overwrite D-14 or the IFC GUID. The manufacturer has a product identifier, the contractor has a procurement record, and the operator will eventually have an asset tag. The translation can describe how these records relate, but must not collapse them into one “door number.” Such collapsing is an attractive shortcut that breaks traceability later.

During construction, a reviewer notes a discrepancy between a site-installed component and the approved detail. That note is evidence of an observed condition, not automatically an instruction to replace the item. The translator preserves the observation and the question assigned to the responsible professional. If a later instruction is issued, it is recorded as a distinct authorised event with its own revision and acknowledgement.

At handover, the verified installed configuration, product documents, final inspection status, warranty limitations and asset tag are connected into the operations register. A target-language maintenance reader should be able to find D-14 and the installed product, understand which manual applies, distinguish an inspection interval from a replacement interval and follow the signed test record. This is the complete-system meaning of translation: the same reality remains navigable after the language changes.

A proper acceptance test uses independent questions. Can the reader identify the controlling plan and its purpose? Can they distinguish nominal from clear dimensions? Can they separate proposed from approved product evidence? Can they locate the actual installed asset? Can they identify unresolved issues without mistaking them for instructions? Can they trace every critical assertion back to an authoritative record? If any answer depends on guessing, the bilingual package is not ready for its declared purpose.

10. Machine translation, extraction and LLM-assisted workflows

Automated translation can help with indexing, candidate terminology, repeated descriptions and draft readability, but it must not be promoted to professional approval. Image extraction can shift callouts, flatten tables and remove a negative sign. OCR may preserve the words while detaching a note from a dimension leader. A language model may expand an abbreviation according to general frequency, then repeat that expansion persuasively across a whole document. These problems require source-aware checks, not just more fluent prompts.

Split data into risk classes before automation. Stable keys and exact geometry should have copy-protection rules and diff tests. Structured bilingual dictionary fields may allow constrained translations. Ordinary explanatory paragraphs can receive broader drafting assistance, provided the technical referents and document status remain visible. High-risk conditional instructions require a competent domain reviewer and a traceable acceptance decision. If you do not know which class a fragment belongs to, do not assume it is harmless text.

A safe prototype process is to extract a source manifest, then generate candidate translations with explicit protected spans and a glossary. The automated stage flags unknown abbreviations, unresolved unit conversions, unpaired table cells and changed identifiers. A second review reconstructs clauses against diagrams and authoritative source data. A third review checks downstream uses, including how exported properties appear in IFC, schedules and operational databases.

Some apparent improvements should be rejected. Do not automatically replace “shall” with a softer expression to sound friendly. Do not rewrite a bare rating into what you believe is an equivalent rating in a different classification system. Do not convert measurements by rounding in a way that changes tolerance or implied precision. Do not fill missing properties with inferred typical values. Do not generate a construction method from a schematic note.

AI can also be used as an adversarial reviewer: ask it to find potentially lost exclusions, status changes, reversed comparisons, ambiguous reference surfaces and one-to-many terminology mappings. Then independently verify every warning; models can invent mismatches as well as miss them. A useful automatic report lists each finding with source location, target location, affected fact, evidence and disposition. It should not claim a system is certified or compliant because a model said no errors were found.

For distributed teams, terminology memory is especially important. Approved translations should include concepts, preferred terms, prohibited equivalents, scope notes, related element classifications, responsible reviewer and effective date. A translation memory alone is insufficient when a familiar word carries a different concept in another discipline. Governance of the glossary matters at least as much as its size.

11. Quality measurements that reveal the right defects

A useful translation quality dashboard measures more than spelling. It checks identifier preservation, unit and sign agreement, table row alignment, numeric referents, revision coverage, status fidelity, conditional obligations, cross-reference reachability, unresolved terminology conflicts and validation of model properties. Each measurement asks a precise question about information integrity.

An identity test compares protected codes, GUIDs, drawing IDs, tags and report numbers between source and output, allowing only documented exceptions. A quantity test compares value, unit, tolerances, inequalities, significant figures and stated reference basis. A relationship test compares which schedule row belongs to which element, which leader line points to which component, and which asset tag links to which installed serial number.

A status test examines whether ‘draft’, ‘proposed’, ‘for review’, ‘approved with comments’, ‘as built’ or analogous project-defined terms retain their control meaning. A source coverage test identifies drawing notes or tables present in one language but absent in the other. A change-impact test checks whether the translation was revalidated after a new authoritative issue. An ambiguity test counts not merely the open items but also their age, owner, affected files and next decision.

These metrics should not be flattened into a misleading universal quality percentage. A package with perfect spelling and one reversed fire-system state can be unfit for use. Risk-based release uses critical defects as gates, separates advisory improvements from blocking errors and makes every acceptance decision attributable. Report known limitations instead of hiding them beneath a high aggregate score.

12. Handover templates for repeatable translation

Template A — Source-control row: document ID; title; discipline; status; revision; purpose; date; controlling source URI or file identity; linked drawing/model IDs; issuing party; recipient; intended target-language use. A row without an explicit purpose should be treated as incomplete for action-bearing documents.

Template B — Protected-field register: field name; original string; data type; unit; reason for non-translation; exception policy; reviewer; test method. Include drawing numbers, IFC GUIDs, equipment tags, controlled property identifiers and technical ratings. Protecting a value does not mean leaving its human-facing explanation untranslated.

Template C — Terminology decision: source expression; concept definition; context; target preferred term; prohibited near-synonym; related discipline; reference; approving specialist; decision date; affected files. A glossary entry should include the reason that an apparently ordinary synonym would alter a component or action.

Template D — Discrepancy ticket: source excerpt; source location; alternative readings; affected identifiers; observed cross-document conflict; risk if unresolved; responsible design owner; response deadline; interim usage restriction; approved resolution; release reference. The ticket is a request for authority, not an opportunity for a translator to redesign the work.

Template E — Bilingual issue record: source manifest; target-language files; format conversions; automated checks; human review signatures; release purpose; approved glossary version; open issues; superseded translation records; recipient acknowledgement. For model deliveries add export settings and the corresponding machine-readable validation results.

Template F — User test: ask a target-language professional to identify the controlling version, affected component, quantity basis, condition, authorising actor and independent verification evidence from the issued materials. Record actual answers, not only whether the reader says the translation feels clear. A failure in one of these questions is a workflow signal to repair the information, not an invitation to blame the user.

13. Terminology map: sixty-three concepts that must not be flattened

This reference map is a concept reminder, not a substitute for a controlled bilingual glossary. Preferred terms depend on discipline, jurisdiction, project contract and standard edition. Use the definitions to ask better questions, then obtain the authoritative project-specific term where necessary.

As-built. Record of the installed or completed condition, not a synonym for a design proposal. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Asset tag. Identifier used by the owner or operator to find a maintainable item. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

BIM. Information processes and models supporting a built asset across relevant lifecycle stages. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Bill of quantities. Commercial list of measured or estimated work items under a defined measurement basis. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Callout. Drawing annotation referring to a specific component, detail or related view. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Clear dimension. Usable unobstructed dimension between defined surfaces or features. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Common data environment. Managed information exchange environment with project-defined status and authorisation. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Commissioning. Documented testing and verification activities for installed systems. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Construction joint. Joint associated with the sequencing or construction of an element. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Coordination issue. Document released for multidisciplinary review rather than automatic construction authority. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Datum. Defined reference used to interpret elevations or measurements. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Detail reference. Pointer to a particular sheet location or explanatory drawing. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Discipline. Authoring or reviewing field, such as structural, architectural or electrical. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Door set. Identified combination of door components whose tested scope must be checked. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Egress. Means or route of leaving an occupied place under defined circumstances. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Element ID. Identifier for an object or component within a project data system. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Expansion joint. Joint addressing movement under a specified design intent. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Fabrication drawing. Information used by the fabricator, subject to the relevant approval chain. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Fall. Gradient or slope in some construction contexts, not automatically a safety incident. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Fire rating. Classification that must retain its test system and stated assembly scope. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Finished floor level. Reference to completed floor surface rather than automatically structural slab top. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Frame size. Dimensions associated with a frame, not necessarily clear passage. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Geometry. Spatial relationships and dimensions that must remain associated with their referents. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Grid reference. Location key in a drawing or model coordinate structure. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

GUID. Globally unique identifier used to preserve identity across digital records. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

IFC. Industry Foundation Classes structured schema for sharing building and infrastructure information. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

IDS. Information delivery specification defining checkable information requirements for certain IFC uses. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Indicative. Qualified information not necessarily confirmed for setting out or construction. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Invert level. Elevation of a specified lower interior flow surface, usually in a drain context. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Issue purpose. Declared use for which a drawing or information container was sent. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Keynote. Coded or numbered explanation tied to a particular drawing feature. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Leader line. Visual line connecting annotation to the intended feature. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Level. Height reference that must be distinguished by datum and surface. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Line load. Force per unit length, a different quantity from force per area. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Manufacturer model. Supplier product designation, not automatically a project element identity. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Material submittal. Submission of proposed product information for a defined review or acceptance. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Minimum. Lower bound; the word must not be smoothed into a typical or average value. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Model view definition. Specified subset and requirements for an IFC exchange workflow. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Nominal. Designation or target that may differ from the actual measured value. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

NCR. Nonconformance report documenting a deviation and its disposition process. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Occurrence. Individual instance of a modelled type or component. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Open issue. Recorded discrepancy not yet resolved by the responsible authority. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Property dictionary. Defined concepts and attributes used to identify model properties consistently. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Property key. Machine-readable reference that should not be replaced with a translated caption. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Provisional. Not final or not guaranteed; retains its contractual or technical qualifier. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

RFI. Request for information, which is a question rather than automatically a site instruction. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Revision. Identified change state governed by the project’s issue and supersession process. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Schedule. Tabular mapping of items, values and identifiers, not a free-text summary. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Schematic. Abstract representation that may not give exact physical routing. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Section. Representation of a cut through an element or arrangement. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Serial number. Unique physical product identifier that must remain byte-accurate. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Shop drawing. Production-oriented information requiring the specified coordination and approval process. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Source hierarchy. Agreed rule for deciding which records control when documents conflict. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Specification. Defined requirements and qualifications for work, materials or performance. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Status code. Project-defined label indicating a document’s permitted information use. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Submittal. Controlled package sent for examination or decision under an agreed workflow. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Superseded. No longer the controlling version for its former defined purpose. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Tolerance. Allowed deviation from a stated reference, distinct from the target value. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Transmittal. Record of which information was sent, by whom, to whom and for what purpose. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Type. Class or product definition shared by multiple occurrences. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

ULS. Ultimate limit state, a structural design concept distinct from serviceability. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Unit. Measurement designation that supplies the dimensional meaning of a number. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Validation. Checking a requirement using evidence, not necessarily professional sign-off. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Variation. Contractual change process that must distinguish proposal from authorised change. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

Witnessed test. Test observed by a party; witness status alone does not establish unconditional approval. When translating, identify whether the expression is a concept, field label, controlled code or ordinary explanation; attach it to the relevant drawing, property or record. An apparent synonym is safe only after the same physical referent and approval boundary have been checked.

14. Twenty-four practical questions and answers

Can ordinary bilingual staff translate construction drawings?

They can help with language and terminology, but a qualified technical reviewer must settle domain meaning, compliance implications and controlled release. Bilingual fluency alone does not authorise changes to design intent. For a live document, attach the source identifier and show the specific evidence used to resolve the issue; broad assurances of accuracy are weaker than an inspectable decision trail.

Should drawing numbers be translated?

Normally preserve controlled sheet IDs, drawing references and revision codes. Translate descriptive titles where useful, while keeping the unchanged identifier visible to every downstream user. For a live document, attach the source identifier and show the specific evidence used to resolve the issue; broad assurances of accuracy are weaker than an inspectable decision trail.

Can I use a machine translator for BIM schedules?

You can use it to generate a candidate human-language field, but protect machine keys, element identity, units and tabular alignment. Compare the resulting model and schedules against the source before release. For a live document, attach the source identifier and show the specific evidence used to resolve the issue; broad assurances of accuracy are weaker than an inspectable decision trail.

What is the difference between an IFC name and a GUID?

A name may be a human-facing description. A GUID is an identity mechanism. Changing display text need not change the GUID; regenerating an ID can disconnect downstream records. For a live document, attach the source identifier and show the specific evidence used to resolve the issue; broad assurances of accuracy are weaker than an inspectable decision trail.

How do you translate building specifications accurately?

Preserve the hierarchy of clauses, conditions, materials, test scope, obligations, exceptions and cross-references. A familiar target-language product category must not be substituted without source approval. For a live document, attach the source identifier and show the specific evidence used to resolve the issue; broad assurances of accuracy are weaker than an inspectable decision trail.

Should abbreviations be expanded?

Only if the discipline and project glossary establish the actual expansion. Many abbreviations carry multiple meanings across drawing families. When unresolved, retain the code and query its owner. For a live document, attach the source identifier and show the specific evidence used to resolve the issue; broad assurances of accuracy are weaker than an inspectable decision trail.

What happens when the drawing and schedule disagree?

Record both source assertions with version and location, then request an authorised clarification. Translation should reveal the conflict, not secretly choose a more attractive number. For a live document, attach the source identifier and show the specific evidence used to resolve the issue; broad assurances of accuracy are weaker than an inspectable decision trail.

Are BIM models always more authoritative than PDFs?

No. Authority depends on the project information hierarchy, issue purpose and approvals. A model may be issued for coordination while a particular drawing or specification controls a different decision. For a live document, attach the source identifier and show the specific evidence used to resolve the issue; broad assurances of accuracy are weaker than an inspectable decision trail.

What is an IFC property dictionary used for?

It helps define and link concepts across software and language views. The translator needs to distinguish a stable property concept or key from a localized display label. For a live document, attach the source identifier and show the specific evidence used to resolve the issue; broad assurances of accuracy are weaker than an inspectable decision trail.

How are revisions handled across languages?

Link each target file to an exact source issue. Perform a change-impact assessment for new revisions and mark superseded target-language copies so they cannot be mistaken for current instructions. For a live document, attach the source identifier and show the specific evidence used to resolve the issue; broad assurances of accuracy are weaker than an inspectable decision trail.

Can one fire rating be automatically converted into another?

Not safely. Ratings depend on applicable classifications, tests and systems. Preserve the original claimed classification and obtain an authorised equivalence judgement if required. For a live document, attach the source identifier and show the specific evidence used to resolve the issue; broad assurances of accuracy are weaker than an inspectable decision trail.

Is a signed inspection report proof of regulatory approval?

Not automatically. It establishes whatever the signed record says under its defined scope. Translate inspection, acceptance, certification and authority decisions as separate events. For a live document, attach the source identifier and show the specific evidence used to resolve the issue; broad assurances of accuracy are weaker than an inspectable decision trail.

How should dimensions be checked?

Compare the number, unit, sign, tolerance, dimension basis, surface reference and attached drawing element. A correct value attached to the wrong component remains a significant error. For a live document, attach the source identifier and show the specific evidence used to resolve the issue; broad assurances of accuracy are weaker than an inspectable decision trail.

What is the most common revision mistake?

Treating an issue purpose such as ‘for review’ as an approval state. A new revision number does not itself authorise building from the sheet. For a live document, attach the source identifier and show the specific evidence used to resolve the issue; broad assurances of accuracy are weaker than an inspectable decision trail.

How can teams localize asset handover records?

Preserve installed asset tag, serial, manufacturer model, maintenance task and test evidence as connected but distinct fields. Localize human instructions without overwriting identity keys. For a live document, attach the source identifier and show the specific evidence used to resolve the issue; broad assurances of accuracy are weaker than an inspectable decision trail.

What if a material specification uses a local designation?

Retain the source designation and governing standard. Do not invent an international equivalent; escalate for a certified equivalence or approved substitution decision. For a live document, attach the source identifier and show the specific evidence used to resolve the issue; broad assurances of accuracy are weaker than an inspectable decision trail.

Can AI verify construction compliance?

Automated checks can find missing properties or inconsistencies, but compliance and engineering responsibility still require the applicable requirements and competent decision-makers. For a live document, attach the source identifier and show the specific evidence used to resolve the issue; broad assurances of accuracy are weaker than an inspectable decision trail.

How do we test whether a translation works?

Ask a target-language user to locate the affected element, identify source authority and version, interpret the constraint and cite verification evidence. Record the response and repair failures. For a live document, attach the source identifier and show the specific evidence used to resolve the issue; broad assurances of accuracy are weaker than an inspectable decision trail.

Should translated drawing notes be inserted directly into CAD?

Only through a controlled process that preserves note placement, references, layers, objects and issue metadata. A text-only export can move context even when wording is accurate. For a live document, attach the source identifier and show the specific evidence used to resolve the issue; broad assurances of accuracy are weaker than an inspectable decision trail.

How do we prevent a multilingual glossary from changing ownership?

Keep the approved language concept glossary specific to this construction workflow and link to broader vocabulary and English resources as educational foundations, not as technical authorities. For a live document, attach the source identifier and show the specific evidence used to resolve the issue; broad assurances of accuracy are weaker than an inspectable decision trail.

Is a technical drawing symbol always universal?

No. Many conventions are shared, but symbols, abbreviations and reference conventions depend on standards, disciplines and drawing legends. Verify the governing key. For a live document, attach the source identifier and show the specific evidence used to resolve the issue; broad assurances of accuracy are weaker than an inspectable decision trail.

How should a translation handle uncertain data?

Preserve ‘estimated’, ‘indicative’, ‘unverified’, ‘pending’ and similar qualifiers. A translated uncertainty is more useful than a fabricated certainty. For a live document, attach the source identifier and show the specific evidence used to resolve the issue; broad assurances of accuracy are weaker than an inspectable decision trail.

Who should approve a technical translation?

The approval structure should match the risk: language reviewers for linguistic clarity, information managers for data integrity and appropriate discipline authorities for technical meaning. For a live document, attach the source identifier and show the specific evidence used to resolve the issue; broad assurances of accuracy are weaker than an inspectable decision trail.

What is the final release criterion?

A target-language user should reach the same supported decision using the same source evidence, while knowing which design decisions remain outside the translation team’s authority. For a live document, attach the source identifier and show the specific evidence used to resolve the issue; broad assurances of accuracy are weaker than an inspectable decision trail.

15. From construction translation to transferable language skill

The most valuable skill in this field is learning how ordinary language connects to precise operational concepts. A word such as “approved” is not a mood, a polite tone or a translation style; it is an evidence-dependent state. A word such as “clear” may be a measurement basis rather than an adjective meaning understandable. A word such as “typical” can restrict the scope of a detail; it need not be an invitation to generalise a rule everywhere.

This is why vocabulary matters, but not as a collection of dictionary equivalents. Technical vocabulary is a system of distinctions tied to observations, measurements, responsibilities and decisions. A learner grows by identifying what two similar words do differently in an actual record. Writing and reading skills then become instruments of checking rather than mere surface fluency.

The method also transfers to laboratory reports, transport records, public procurement, medicine, safety notices and digital software. Protect identity and evidence; keep scope and qualifiers; distinguish measured from inferred values; attach claims to the right object; require professional authority where the translation could alter a consequential action. These habits scale beyond the present specialty without making the present specialty the owner of every other field.

This article therefore makes a deliberate boundary choice. It does not redefine the general translation system, teach an entire vocabulary curriculum or attempt to own all English-language grammar. It contributes a specialized construction-and-BIM case system that readers can route back to the larger framework when they need foundational language methods.

16. References, next routes and editorial limits

Return to the root: Master Art of Translation | The Complete System for Moving Meaning Between Languages. Read the separate short applied guide: Translate Construction Drawings, Specifications and Site Instructions Without Changing Dimensions or Scope. For transferable word-learning, use the protected Vocabulary Learning Hub; for sentence structure, explanation and English learning routes, use the protected How English Works. These are outward educational routes, not claims that those pages govern an engineering design.

Primary technical references: ISO 19650-1:2018; ISO 19650-2:2018; ISO 16739-1:2024 (IFC); ISO 23386:2020 (properties); ISO 12006-2:2015 (classification framework); ISO 128-1:2020 (technical drawings); ISO 21542:2021 (accessibility); ISO 19650-6:2025 (health and safety information); buildingSMART IDS v1.0 announcement (2024). Links checked 30 October 2026; standards and regulations may be amended after this date.

Editorial disclosure: fictional worked examples explain translation diagnostics rather than provide construction drawings, buildable details, design calculations, product approvals, jurisdictional fire classifications, safe working methods or legally enforceable interpretations. For real projects follow the current standards and local authorities, obtain the source documents and engage the responsible licensed or otherwise competent professional. Translate only what the record proves, preserve what is unresolved and never substitute fluency for authority.

Master Art of Translation | The Construction Drawings & BIM Information Translation System — How Plans, Specifications, Quantities, Revisions and Asset Records Keep Their Meaning