Master Art of Translation | The Clinical Trials & Informed Consent Translation System — How Protocols, Participant Rights, Safety Reports and Outcomes Keep Their Meaning

Clinical trial translation: the answer people need first

Clinical trial translation is the specialised work of translating informed consent forms, research protocols, participant information sheets, study recruitment materials, clinical study reports and adverse-event communications without changing what a participant, investigator, sponsor or reviewer can understand and decide. People searching for medical research translation, informed consent form translation, clinical trial document translation, GCP translation, patient-facing clinical research language or multilingual trial safety reporting need a practical answer: preserve the difference between research and care, possibility and certainty, eligibility and enrolment, a suspected safety event and a confirmed causal relationship, and an informed choice and a signature.

A trustworthy translation workflow protects not merely scientific vocabulary but participant autonomy and the evidentiary record. It begins with the approved source and its intended audience, identifies each decision-critical claim, controls terms and document versions, translates for comprehension, checks every number and condition, tests the material with qualified linguistic and subject-matter reviewers, and releases only the approved version. The same principle governs consent discussions, questionnaires, investigator instructions, protocol amendments, data collection forms and public trial-register summaries: language must not make the research sound safer, more certain, more beneficial, more obligatory or more successful than the source supports.

This world-facing master guide explains how clinical trial translation and multilingual informed consent actually work, why a fluent sentence can still be misleading, how to build a repeatable review system, and how to diagnose errors with fictional worked examples. It distinguishes professional translation from interpreting, educational quality checks from ethics approval, protocol terminology from patient-friendly explanation, and local regulatory requirements from general good practice. No single country’s consent rules govern the world; the controlling protocol, ethics approvals and applicable law must be checked for each real study. This page teaches how to preserve meaning; it does not supply medical advice, determine eligibility, create consent, or certify regulatory compliance.

For the general process, return to Master Art of Translation | The Complete System for Moving Meaning Between Languages. This new specialist child owns the narrower clinical-research translation problem, and leaves existing translation, vocabulary and English-learning pages in their original roles.

Choose the right route through this guide

If you are a prospective participant, begin with the section on understandable consent and the examples about rights, risk, random allocation and withdrawal. If you are a translator, use the invariant checklist, document map and diagnostics. If you manage a trial, start with governance, source-version control and review gates. If you review a protocol or registry entry, prioritise eligibility, endpoints, safety classifications and statistical uncertainty. If you teach language, study how ordinary words such as may, likely, associated, required, and optional change decisions.

The hidden problem: language is part of the research intervention

Imagine two people reviewing what is ostensibly the same trial invitation. One reads that joining a study may offer an opportunity to receive an investigational treatment. The other reads that participation will provide a new treatment. The second version can be grammatically smooth and factually false. It turns a possibility into a guarantee. A participant may accept a burden or risk for a benefit that was never promised. That is why the unit of quality in research translation is not the sentence alone. It is the decision made possible by the sentence.

A second problem is the boundary between clinical care and research. A treating clinician may recommend care on an individual basis; a trial protocol tests a question under pre-specified conditions. Words such as treatment, standard of care, experimental, assigned and withdraw are therefore not stylistic decorations. They connect to different actions, obligations and expectations. The translator must preserve those relationships even when the target language has a less compact vocabulary.

A third problem is document authority. A sponsor may revise the protocol, a research ethics committee may approve a consent form, a trial registry may carry a public summary, and investigators may receive local operating instructions. These texts are connected but are not interchangeable. An elegant translation of a superseded consent form is still an unapproved version. A casual replacement of an official term in a structured case-report form can break data mapping. A correct public summary can be misleading if it silently imports a stronger efficacy claim from promotional material.

A fourth problem is that speaking and reading are not the same. A competent interpreter helps a conversation happen; a verified written translation gives people an accurate document to study and revisit. A participant may speak one language fluently and read another with difficulty, or may need accessible formats. A signature demonstrates a recorded step, not that every important distinction has been understood. Translation quality therefore includes comprehension testing, a chance to ask questions, and an appropriate route back to qualified trial personnel.

A fifth problem is changing evidence. A study can move from recruiting to suspended, an adverse event may initially be reported before later assessment, a protocol may change its visit schedule, and an outcome may be updated after statistical analysis. Translation must preserve the specific state and date of each record. Do not use later knowledge to make earlier language appear prescient; do not hide corrections by quietly replacing a posted version.

The translation mechanism: twenty meaning invariants

1. Research versus routine care

Preserve: Identify every place the source says research, investigational, routine clinical care, standard treatment, or medical recommendation. A reader should know which procedures are performed because of the study and which belong to normal care. The stable object is the participant’s or reviewer’s interpretation, not the original word order. Translate the whole claim, including qualifications, responsibility, exceptions, scope and time. If a specialist must consult the underlying record to know what a phrase means, that consultation is part of the translation task rather than a reason to guess.

Review test: Compare the participant information sheet, protocol schedule and investigator explanation; flag any statement that sounds like a clinician has promised a personalised benefit. Record the source segment, agreed term, open question, responsible owner and answer. Never declare a phrase approved merely because a bilingual reviewer found it idiomatic; ask whether a person reading only the target language would infer the same rights, constraints and evidence as a person reading the controlled original.

2. Voluntary choice

Preserve: Preserve the right to decline participation, decide without pressure, and ask questions. Invitation, request, condition, obligation and permission must remain distinct. The stable object is the participant’s or reviewer’s interpretation, not the original word order. Translate the whole claim, including qualifications, responsibility, exceptions, scope and time. If a specialist must consult the underlying record to know what a phrase means, that consultation is part of the translation task rather than a reason to guess.

Review test: Read the translation as a cautious participant: does any phrase sound like enrolment is required to obtain ordinary care? Record the source segment, agreed term, open question, responsible owner and answer. Never declare a phrase approved merely because a bilingual reviewer found it idiomatic; ask whether a person reading only the target language would infer the same rights, constraints and evidence as a person reading the controlled original.

3. Withdrawal and its limits

Preserve: Keep the right to stop study participation separate from limits on removing data already collected, stopping a study drug safely, or obligations specified by the approved protocol. Never invent a promise about data erasure. The stable object is the participant’s or reviewer’s interpretation, not the original word order. Translate the whole claim, including qualifications, responsibility, exceptions, scope and time. If a specialist must consult the underlying record to know what a phrase means, that consultation is part of the translation task rather than a reason to guess.

Review test: Check which actual source clause explains what happens after withdrawal, and preserve its carefully bounded scope. Record the source segment, agreed term, open question, responsible owner and answer. Never declare a phrase approved merely because a bilingual reviewer found it idiomatic; ask whether a person reading only the target language would infer the same rights, constraints and evidence as a person reading the controlled original.

4. Investigational status

Preserve: Maintain investigational, experimental, approved, authorised, licensed, unapproved, placebo and standard-of-care distinctions according to the source and relevant jurisdiction. The stable object is the participant’s or reviewer’s interpretation, not the original word order. Translate the whole claim, including qualifications, responsibility, exceptions, scope and time. If a specialist must consult the underlying record to know what a phrase means, that consultation is part of the translation task rather than a reason to guess.

Review test: Ask a reviewer to identify what is being studied, not merely what the product is called. Record the source segment, agreed term, open question, responsible owner and answer. Never declare a phrase approved merely because a bilingual reviewer found it idiomatic; ask whether a person reading only the target language would infer the same rights, constraints and evidence as a person reading the controlled original.

5. Possibility, frequency and uncertainty

Preserve: Retain may, might, could, unknown, sometimes, common, uncommon, estimated, not established and cannot be ruled out without silently strengthening or weakening them. The stable object is the participant’s or reviewer’s interpretation, not the original word order. Translate the whole claim, including qualifications, responsibility, exceptions, scope and time. If a specialist must consult the underlying record to know what a phrase means, that consultation is part of the translation task rather than a reason to guess.

Review test: Compare modality and evidence strength line by line; a smooth sentence may alter the underlying probability statement. Record the source segment, agreed term, open question, responsible owner and answer. Never declare a phrase approved merely because a bilingual reviewer found it idiomatic; ask whether a person reading only the target language would infer the same rights, constraints and evidence as a person reading the controlled original.

6. Randomisation and allocation

Preserve: Distinguish random assignment from random sampling, treatment choice from allocation concealment, and assignment probability from benefit probability. The stable object is the participant’s or reviewer’s interpretation, not the original word order. Translate the whole claim, including qualifications, responsibility, exceptions, scope and time. If a specialist must consult the underlying record to know what a phrase means, that consultation is part of the translation task rather than a reason to guess.

Review test: Retain ratios, strata and descriptions of who can choose or know an assignment without guessing missing details. Record the source segment, agreed term, open question, responsible owner and answer. Never declare a phrase approved merely because a bilingual reviewer found it idiomatic; ask whether a person reading only the target language would infer the same rights, constraints and evidence as a person reading the controlled original.

7. Blinding and masking

Preserve: Preserve who is masked, which information is masked, and when unmasking may occur. Do not mistake single-blind for a universal configuration. The stable object is the participant’s or reviewer’s interpretation, not the original word order. Translate the whole claim, including qualifications, responsibility, exceptions, scope and time. If a specialist must consult the underlying record to know what a phrase means, that consultation is part of the translation task rather than a reason to guess.

Review test: Match terms to the protocol’s actual groups and roles rather than relying on generic textbook shorthand. Record the source segment, agreed term, open question, responsible owner and answer. Never declare a phrase approved merely because a bilingual reviewer found it idiomatic; ask whether a person reading only the target language would infer the same rights, constraints and evidence as a person reading the controlled original.

8. Eligibility and screening

Preserve: Separate pre-screening interest, screening assessment, eligible status, enrolled status and randomised status. Exclusion criteria are not automatically diagnoses. The stable object is the participant’s or reviewer’s interpretation, not the original word order. Translate the whole claim, including qualifications, responsibility, exceptions, scope and time. If a specialist must consult the underlying record to know what a phrase means, that consultation is part of the translation task rather than a reason to guess.

Review test: Trace status labels through screening logs, study invitations and registry entries. Record the source segment, agreed term, open question, responsible owner and answer. Never declare a phrase approved merely because a bilingual reviewer found it idiomatic; ask whether a person reading only the target language would infer the same rights, constraints and evidence as a person reading the controlled original.

9. Population and denominator

Preserve: Keep track of who the analysis counts: enrolled, randomised, treated, exposed, evaluable, completed or included in a particular analysis. The stable object is the participant’s or reviewer’s interpretation, not the original word order. Translate the whole claim, including qualifications, responsibility, exceptions, scope and time. If a specialist must consult the underlying record to know what a phrase means, that consultation is part of the translation task rather than a reason to guess.

Review test: Attach every percentage to its denominator and population; do not infer a missing group definition. Record the source segment, agreed term, open question, responsible owner and answer. Never declare a phrase approved merely because a bilingual reviewer found it idiomatic; ask whether a person reading only the target language would infer the same rights, constraints and evidence as a person reading the controlled original.

10. Interventions and comparator

Preserve: Preserve intervention, comparator, sham, placebo, routine care, rescue therapy and concomitant medication as distinct concepts. The stable object is the participant’s or reviewer’s interpretation, not the original word order. Translate the whole claim, including qualifications, responsibility, exceptions, scope and time. If a specialist must consult the underlying record to know what a phrase means, that consultation is part of the translation task rather than a reason to guess.

Review test: Check the intervention schedule and whether the comparator is active, inactive or procedural. Record the source segment, agreed term, open question, responsible owner and answer. Never declare a phrase approved merely because a bilingual reviewer found it idiomatic; ask whether a person reading only the target language would infer the same rights, constraints and evidence as a person reading the controlled original.

11. Doses, units and timing

Preserve: Protect units, dose per administration, total daily dose, route, duration, cycle length, windows and conditional instructions exactly as supplied. The stable object is the participant’s or reviewer’s interpretation, not the original word order. Translate the whole claim, including qualifications, responsibility, exceptions, scope and time. If a specialist must consult the underlying record to know what a phrase means, that consultation is part of the translation task rather than a reason to guess.

Review test: No translated passage may function as a replacement dosing instruction; escalate any mismatch with an approved protocol. Record the source segment, agreed term, open question, responsible owner and answer. Never declare a phrase approved merely because a bilingual reviewer found it idiomatic; ask whether a person reading only the target language would infer the same rights, constraints and evidence as a person reading the controlled original.

12. Visits and procedures

Preserve: Distinguish planned visit, optional follow-up, unscheduled assessment, telephone contact, procedure, sample collection and remote report. The stable object is the participant’s or reviewer’s interpretation, not the original word order. Translate the whole claim, including qualifications, responsibility, exceptions, scope and time. If a specialist must consult the underlying record to know what a phrase means, that consultation is part of the translation task rather than a reason to guess.

Review test: Compare the schedule-of-activities grid to each participant sentence and mark changes in burden or timing. Record the source segment, agreed term, open question, responsible owner and answer. Never declare a phrase approved merely because a bilingual reviewer found it idiomatic; ask whether a person reading only the target language would infer the same rights, constraints and evidence as a person reading the controlled original.

13. Adverse event classifications

Preserve: Keep event, serious adverse event, adverse reaction, suspected reaction, severity grade, seriousness and causality assessment separate. The stable object is the participant’s or reviewer’s interpretation, not the original word order. Translate the whole claim, including qualifications, responsibility, exceptions, scope and time. If a specialist must consult the underlying record to know what a phrase means, that consultation is part of the translation task rather than a reason to guess.

Review test: Confirm whether the source labels an occurrence, a determination, a coding category or a reporting obligation. Record the source segment, agreed term, open question, responsible owner and answer. Never declare a phrase approved merely because a bilingual reviewer found it idiomatic; ask whether a person reading only the target language would infer the same rights, constraints and evidence as a person reading the controlled original.

14. Safety response and contact

Preserve: Preserve what to report, to whom, when and through which approved channel. Do not add emergency advice or delay an urgent reporting instruction. The stable object is the participant’s or reviewer’s interpretation, not the original word order. Translate the whole claim, including qualifications, responsibility, exceptions, scope and time. If a specialist must consult the underlying record to know what a phrase means, that consultation is part of the translation task rather than a reason to guess.

Review test: Check real contact details in approved local materials and distinguish example placeholders from operational data. Record the source segment, agreed term, open question, responsible owner and answer. Never declare a phrase approved merely because a bilingual reviewer found it idiomatic; ask whether a person reading only the target language would infer the same rights, constraints and evidence as a person reading the controlled original.

15. Outcome versus result

Preserve: Maintain primary, secondary, exploratory, surrogate, endpoint, measurement, outcome and estimate as different terms. The stable object is the participant’s or reviewer’s interpretation, not the original word order. Translate the whole claim, including qualifications, responsibility, exceptions, scope and time. If a specialist must consult the underlying record to know what a phrase means, that consultation is part of the translation task rather than a reason to guess.

Review test: Verify that an endpoint definition in the protocol is not quietly rewritten using the study’s eventual result. Record the source segment, agreed term, open question, responsible owner and answer. Never declare a phrase approved merely because a bilingual reviewer found it idiomatic; ask whether a person reading only the target language would infer the same rights, constraints and evidence as a person reading the controlled original.

16. Statistical uncertainty

Preserve: Do not convert an association into causation, statistical significance into clinical importance, or an interval estimate into a guarantee. The stable object is the participant’s or reviewer’s interpretation, not the original word order. Translate the whole claim, including qualifications, responsibility, exceptions, scope and time. If a specialist must consult the underlying record to know what a phrase means, that consultation is part of the translation task rather than a reason to guess.

Review test: Keep effect size, precision, interval boundaries, population and analysis assumptions linked. Record the source segment, agreed term, open question, responsible owner and answer. Never declare a phrase approved merely because a bilingual reviewer found it idiomatic; ask whether a person reading only the target language would infer the same rights, constraints and evidence as a person reading the controlled original.

17. Participant burden and compensation

Preserve: Protect reimbursement, compensation, payment, inducement, travel cost, injury arrangements and financial responsibility distinctions. The stable object is the participant’s or reviewer’s interpretation, not the original word order. Translate the whole claim, including qualifications, responsibility, exceptions, scope and time. If a specialist must consult the underlying record to know what a phrase means, that consultation is part of the translation task rather than a reason to guess.

Review test: Ask whether a participant might believe payment is guaranteed, conditional, punitive or tied to a favourable result. Record the source segment, agreed term, open question, responsible owner and answer. Never declare a phrase approved merely because a bilingual reviewer found it idiomatic; ask whether a person reading only the target language would infer the same rights, constraints and evidence as a person reading the controlled original.

18. Privacy and future use

Preserve: Distinguish anonymised, pseudonymised, coded, identifiable, de-identified, access-restricted, storage and reuse consent. The stable object is the participant’s or reviewer’s interpretation, not the original word order. Translate the whole claim, including qualifications, responsibility, exceptions, scope and time. If a specialist must consult the underlying record to know what a phrase means, that consultation is part of the translation task rather than a reason to guess.

Review test: Inspect the data-flow and retention description; never promise absolute anonymity when the source describes coding or controlled access. Record the source segment, agreed term, open question, responsible owner and answer. Never declare a phrase approved merely because a bilingual reviewer found it idiomatic; ask whether a person reading only the target language would infer the same rights, constraints and evidence as a person reading the controlled original.

19. Protocol version and provenance

Preserve: Track protocol number, amendment, document title, language, country or region where necessary, approval date, translation version and release state. The stable object is the participant’s or reviewer’s interpretation, not the original word order. Translate the whole claim, including qualifications, responsibility, exceptions, scope and time. If a specialist must consult the underlying record to know what a phrase means, that consultation is part of the translation task rather than a reason to guess.

Review test: Reject a text that has excellent language but cannot be matched to an authorised source and effective version. Record the source segment, agreed term, open question, responsible owner and answer. Never declare a phrase approved merely because a bilingual reviewer found it idiomatic; ask whether a person reading only the target language would infer the same rights, constraints and evidence as a person reading the controlled original.

20. Audience and comprehension

Preserve: Translate technical terms consistently while explaining decisions in language readers can understand. A glossary must support, not override, the approved source. The stable object is the participant’s or reviewer’s interpretation, not the original word order. Translate the whole claim, including qualifications, responsibility, exceptions, scope and time. If a specialist must consult the underlying record to know what a phrase means, that consultation is part of the translation task rather than a reason to guess.

Review test: Use teach-back or other suitable checks led by trial professionals where required; record difficulties and correct the controlled content. Record the source segment, agreed term, open question, responsible owner and answer. Never declare a phrase approved merely because a bilingual reviewer found it idiomatic; ask whether a person reading only the target language would infer the same rights, constraints and evidence as a person reading the controlled original.

The repeatable end-to-end method: from request to controlled release

Step 1: Define the decision at stake

Write down whether the text informs consent, controls a procedure, records data, reports safety or communicates results. Each requires a different level of specialist review. The deliverable for this step is a traceable decision: what source material was examined, which terms or values were held invariant, what could legitimately be adapted for the audience, and what remains unresolved. Keeping this evidence makes it possible to explain a correction without accusing a translator or pretending that a fluent draft is final.

For a useful acceptance check, hand the proposed wording to someone who has not seen the source. Ask them to explain the practical consequence in their own words. Compare that explanation with the controlled source and the protocol’s current status. If their answer changes an eligibility condition, a participant right, a timing obligation, a safety classification or a claim about evidence, reopen the segment. This process is deliberately slower at the points where an error would matter most.

Step 2: Map every source document

List the master protocol, amendments, consent sheets, assent forms, questionnaires, investigator instructions, registry entries, emails and study reports; preserve their IDs. The deliverable for this step is a traceable decision: what source material was examined, which terms or values were held invariant, what could legitimately be adapted for the audience, and what remains unresolved. Keeping this evidence makes it possible to explain a correction without accusing a translator or pretending that a fluent draft is final.

Step 3: Name the controlling authority

Document the version approved or otherwise authorised for the relevant use. Separate an institutional template from the actual approval of study-specific material. The deliverable for this step is a traceable decision: what source material was examined, which terms or values were held invariant, what could legitimately be adapted for the audience, and what remains unresolved. Keeping this evidence makes it possible to explain a correction without accusing a translator or pretending that a fluent draft is final.

Step 4: Assign risk by sentence

Mark consent rights, dosing units, emergency contact, eligibility rules, adverse-event words, endpoint definitions and statistics as meaning-critical. The deliverable for this step is a traceable decision: what source material was examined, which terms or values were held invariant, what could legitimately be adapted for the audience, and what remains unresolved. Keeping this evidence makes it possible to explain a correction without accusing a translator or pretending that a fluent draft is final.

Step 5: Build a concept glossary

Create entries for randomised, blinded, cohort, comparator, placebo, primary endpoint, treatment-emergent event, relatedness and withdrawal with context. The deliverable for this step is a traceable decision: what source material was examined, which terms or values were held invariant, what could legitimately be adapted for the audience, and what remains unresolved. Keeping this evidence makes it possible to explain a correction without accusing a translator or pretending that a fluent draft is final.

Step 6: Specify the target reader

Describe a participant’s expected literacy, familiarity with medical research, accessibility needs and language of discussion without stereotyping the community. The deliverable for this step is a traceable decision: what source material was examined, which terms or values were held invariant, what could legitimately be adapted for the audience, and what remains unresolved. Keeping this evidence makes it possible to explain a correction without accusing a translator or pretending that a fluent draft is final.

Step 7: Separate translation and interpreting

Agree who will translate written materials and who will interpret discussions; document their competence and avoid pretending one substitutes for the other. The deliverable for this step is a traceable decision: what source material was examined, which terms or values were held invariant, what could legitimately be adapted for the audience, and what remains unresolved. Keeping this evidence makes it possible to explain a correction without accusing a translator or pretending that a fluent draft is final.

Step 8: Prepare the translator brief

Give the role, source status, audience, medical background, restricted terms, allowed abbreviations, protected identifiers and escalation contacts. The deliverable for this step is a traceable decision: what source material was examined, which terms or values were held invariant, what could legitimately be adapted for the audience, and what remains unresolved. Keeping this evidence makes it possible to explain a correction without accusing a translator or pretending that a fluent draft is final.

Step 9: Translate claims rather than words

Ask what a sentence permits, predicts, requires, denies or leaves unknown. Preserve logical operators, conditions, exceptions and uncertainty. The deliverable for this step is a traceable decision: what source material was examined, which terms or values were held invariant, what could legitimately be adapted for the audience, and what remains unresolved. Keeping this evidence makes it possible to explain a correction without accusing a translator or pretending that a fluent draft is final.

Step 10: Freeze structured values

Protect dates, numbers, units, code values, arm names, visit IDs, population counts, questionnaire anchors, randomisation ratios and record IDs. The deliverable for this step is a traceable decision: what source material was examined, which terms or values were held invariant, what could legitimately be adapted for the audience, and what remains unresolved. Keeping this evidence makes it possible to explain a correction without accusing a translator or pretending that a fluent draft is final.

Step 11: Run numerical comparison

Perform an independent pass through every digit, sign, comparator, percentage, denominator, confidence interval, date and study duration. The deliverable for this step is a traceable decision: what source material was examined, which terms or values were held invariant, what could legitimately be adapted for the audience, and what remains unresolved. Keeping this evidence makes it possible to explain a correction without accusing a translator or pretending that a fluent draft is final.

Step 12: Review all participant decisions

Ask whether the target-language reader can understand the purpose, foreseeable risk, alternatives, burden, voluntary choice and contact path. The deliverable for this step is a traceable decision: what source material was examined, which terms or values were held invariant, what could legitimately be adapted for the audience, and what remains unresolved. Keeping this evidence makes it possible to explain a correction without accusing a translator or pretending that a fluent draft is final.

Step 13: Run a qualified language revision

Use a second competent reviewer where the project’s standard requires it; resolve differences through a documented decision, not a popularity vote. The deliverable for this step is a traceable decision: what source material was examined, which terms or values were held invariant, what could legitimately be adapted for the audience, and what remains unresolved. Keeping this evidence makes it possible to explain a correction without accusing a translator or pretending that a fluent draft is final.

Step 14: Obtain scientific verification

Have an appropriately qualified subject-matter reviewer check research concepts, intervention descriptions, endpoints and safety vocabulary. The deliverable for this step is a traceable decision: what source material was examined, which terms or values were held invariant, what could legitimately be adapted for the audience, and what remains unresolved. Keeping this evidence makes it possible to explain a correction without accusing a translator or pretending that a fluent draft is final.

Step 15: Check the real layout

Review mobile screens, PDFs, tables, callouts, footnotes and signatures. A missing negation or hidden qualifier can be a layout error. The deliverable for this step is a traceable decision: what source material was examined, which terms or values were held invariant, what could legitimately be adapted for the audience, and what remains unresolved. Keeping this evidence makes it possible to explain a correction without accusing a translator or pretending that a fluent draft is final.

Step 16: Test accessible comprehension

Evaluate whether readers can distinguish the decisions that matter, not just repeat a technical term; preserve approved, meaningful wording. The deliverable for this step is a traceable decision: what source material was examined, which terms or values were held invariant, what could legitimately be adapted for the audience, and what remains unresolved. Keeping this evidence makes it possible to explain a correction without accusing a translator or pretending that a fluent draft is final.

Step 17: Reconcile parallel languages

Compare all target versions against the same controlling source, not merely against one another. Record why any language uses extra explanatory words. The deliverable for this step is a traceable decision: what source material was examined, which terms or values were held invariant, what could legitimately be adapted for the audience, and what remains unresolved. Keeping this evidence makes it possible to explain a correction without accusing a translator or pretending that a fluent draft is final.

Step 18: Resolve source ambiguity

If the source itself is unclear, ask its owner to clarify in the controlled record rather than inventing a translation that masks a defect. The deliverable for this step is a traceable decision: what source material was examined, which terms or values were held invariant, what could legitimately be adapted for the audience, and what remains unresolved. Keeping this evidence makes it possible to explain a correction without accusing a translator or pretending that a fluent draft is final.

Step 19: Complete approval and release

Match the translation package to required sponsor, investigator and ethics/regulatory processes, as applicable. Translation approval is not automatically ethics approval. The deliverable for this step is a traceable decision: what source material was examined, which terms or values were held invariant, what could legitimately be adapted for the audience, and what remains unresolved. Keeping this evidence makes it possible to explain a correction without accusing a translator or pretending that a fluent draft is final.

Step 20: Distribute with version control

Publish or send only authorised versions, with effective date and audience. Remove obsolete copies from live distribution while preserving audit history. The deliverable for this step is a traceable decision: what source material was examined, which terms or values were held invariant, what could legitimately be adapted for the audience, and what remains unresolved. Keeping this evidence makes it possible to explain a correction without accusing a translator or pretending that a fluent draft is final.

Step 21: Handle changes and reconsent

When a source amendment changes a relevant decision, run a delta comparison and have responsible personnel determine whether retranslation, reapproval or renewed consent is required. The deliverable for this step is a traceable decision: what source material was examined, which terms or values were held invariant, what could legitimately be adapted for the audience, and what remains unresolved. Keeping this evidence makes it possible to explain a correction without accusing a translator or pretending that a fluent draft is final.

Step 22: Learn from incidents

Log comprehension complaints, wording corrections and downstream data mismatches. Improve the glossary and review rules without silently rewriting historical records. The deliverable for this step is a traceable decision: what source material was examined, which terms or values were held invariant, what could legitimately be adapted for the audience, and what remains unresolved. Keeping this evidence makes it possible to explain a correction without accusing a translator or pretending that a fluent draft is final.

Failure modes: where apparently good translation breaks

1. The guaranteed-benefit drift

Failure: A possibility of benefit becomes a promised improvement. This can survive ordinary spell-checking and even a fluent bilingual read because the missing distinction sits in the interaction between wording, protocol context, presentation and the reader’s likely decision. Ask what the affected person would do differently if they trusted this version rather than the original.

Repair: Preserve uncertain benefit wording, and ask if the text changes willingness to participate. Then update the termbase or quality checklist with a narrowly scoped rule; assign the source owner any scientific ambiguity. Record whether a released version needs withdrawal or correction. A robust fix controls both the sentence and the document lifecycle, rather than merely polishing a phrase and declaring the problem solved.

2. Research becomes treatment

Failure: An investigational procedure is described as recommended ordinary care. This can survive ordinary spell-checking and even a fluent bilingual read because the missing distinction sits in the interaction between wording, protocol context, presentation and the reader’s likely decision. Ask what the affected person would do differently if they trusted this version rather than the original.

Repair: Separate the study’s purpose from an individual clinician’s treatment plan. Then update the termbase or quality checklist with a narrowly scoped rule; assign the source owner any scientific ambiguity. Record whether a released version needs withdrawal or correction. A robust fix controls both the sentence and the document lifecycle, rather than merely polishing a phrase and declaring the problem solved.

3. Voluntary becomes necessary

Failure: A polite invitation becomes a condition for access to services. This can survive ordinary spell-checking and even a fluent bilingual read because the missing distinction sits in the interaction between wording, protocol context, presentation and the reader’s likely decision. Ask what the affected person would do differently if they trusted this version rather than the original.

Repair: Maintain the source’s choice language and any approved care alternatives. Then update the termbase or quality checklist with a narrowly scoped rule; assign the source owner any scientific ambiguity. Record whether a released version needs withdrawal or correction. A robust fix controls both the sentence and the document lifecycle, rather than merely polishing a phrase and declaring the problem solved.

4. Placebo becomes no care

Failure: A comparator is casually described as withholding all treatment. This can survive ordinary spell-checking and even a fluent bilingual read because the missing distinction sits in the interaction between wording, protocol context, presentation and the reader’s likely decision. Ask what the affected person would do differently if they trusted this version rather than the original.

Repair: Explain the actual design only to the extent the approved source describes it. Then update the termbase or quality checklist with a narrowly scoped rule; assign the source owner any scientific ambiguity. Record whether a released version needs withdrawal or correction. A robust fix controls both the sentence and the document lifecycle, rather than merely polishing a phrase and declaring the problem solved.

5. Assignment becomes choice

Failure: Random allocation is translated as selecting one’s preferred group. This can survive ordinary spell-checking and even a fluent bilingual read because the missing distinction sits in the interaction between wording, protocol context, presentation and the reader’s likely decision. Ask what the affected person would do differently if they trusted this version rather than the original.

Repair: Check the allocation mechanism and the source’s stated probabilities. Then update the termbase or quality checklist with a narrowly scoped rule; assign the source owner any scientific ambiguity. Record whether a released version needs withdrawal or correction. A robust fix controls both the sentence and the document lifecycle, rather than merely polishing a phrase and declaring the problem solved.

6. Screened becomes enrolled

Failure: A screening appointment is described as confirmed participation. This can survive ordinary spell-checking and even a fluent bilingual read because the missing distinction sits in the interaction between wording, protocol context, presentation and the reader’s likely decision. Ask what the affected person would do differently if they trusted this version rather than the original.

Repair: Preserve each trial-status boundary; screen failures are not participants in every analysis. Then update the termbase or quality checklist with a narrowly scoped rule; assign the source owner any scientific ambiguity. Record whether a released version needs withdrawal or correction. A robust fix controls both the sentence and the document lifecycle, rather than merely polishing a phrase and declaring the problem solved.

7. Serious becomes severe

Failure: Regulatory seriousness is confused with clinical intensity. This can survive ordinary spell-checking and even a fluent bilingual read because the missing distinction sits in the interaction between wording, protocol context, presentation and the reader’s likely decision. Ask what the affected person would do differently if they trusted this version rather than the original.

Repair: Check the intended safety category and do not assume a one-to-one synonym. Then update the termbase or quality checklist with a narrowly scoped rule; assign the source owner any scientific ambiguity. Record whether a released version needs withdrawal or correction. A robust fix controls both the sentence and the document lifecycle, rather than merely polishing a phrase and declaring the problem solved.

8. Relatedness becomes causation

Failure: An event reported after treatment is declared caused by treatment. This can survive ordinary spell-checking and even a fluent bilingual read because the missing distinction sits in the interaction between wording, protocol context, presentation and the reader’s likely decision. Ask what the affected person would do differently if they trusted this version rather than the original.

Repair: Keep temporal sequence and investigator causality assessment separate. Then update the termbase or quality checklist with a narrowly scoped rule; assign the source owner any scientific ambiguity. Record whether a released version needs withdrawal or correction. A robust fix controls both the sentence and the document lifecycle, rather than merely polishing a phrase and declaring the problem solved.

9. No evidence becomes evidence of no effect

Failure: An inconclusive analysis is presented as proof of equivalence or safety. This can survive ordinary spell-checking and even a fluent bilingual read because the missing distinction sits in the interaction between wording, protocol context, presentation and the reader’s likely decision. Ask what the affected person would do differently if they trusted this version rather than the original.

Repair: Preserve limitations, confidence intervals and the claim actually tested. Then update the termbase or quality checklist with a narrowly scoped rule; assign the source owner any scientific ambiguity. Record whether a released version needs withdrawal or correction. A robust fix controls both the sentence and the document lifecycle, rather than merely polishing a phrase and declaring the problem solved.

10. Risk is hidden in a footnote

Failure: A visible headline sounds reassuring while a serious qualifier is buried. This can survive ordinary spell-checking and even a fluent bilingual read because the missing distinction sits in the interaction between wording, protocol context, presentation and the reader’s likely decision. Ask what the affected person would do differently if they trusted this version rather than the original.

Repair: Check information hierarchy and participant-facing comprehension. Then update the termbase or quality checklist with a narrowly scoped rule; assign the source owner any scientific ambiguity. Record whether a released version needs withdrawal or correction. A robust fix controls both the sentence and the document lifecycle, rather than merely polishing a phrase and declaring the problem solved.

11. Withdraw means erase

Failure: Ending participation is rendered as erasure of all collected records. This can survive ordinary spell-checking and even a fluent bilingual read because the missing distinction sits in the interaction between wording, protocol context, presentation and the reader’s likely decision. Ask what the affected person would do differently if they trusted this version rather than the original.

Repair: Explain data retention only as the controlled source permits. Then update the termbase or quality checklist with a narrowly scoped rule; assign the source owner any scientific ambiguity. Record whether a released version needs withdrawal or correction. A robust fix controls both the sentence and the document lifecycle, rather than merely polishing a phrase and declaring the problem solved.

12. Optional becomes mandatory

Failure: A questionnaire item or specimen donation is made obligatory. This can survive ordinary spell-checking and even a fluent bilingual read because the missing distinction sits in the interaction between wording, protocol context, presentation and the reader’s likely decision. Ask what the affected person would do differently if they trusted this version rather than the original.

Repair: Preserve selection, refusal and consequences separately. Then update the termbase or quality checklist with a narrowly scoped rule; assign the source owner any scientific ambiguity. Record whether a released version needs withdrawal or correction. A robust fix controls both the sentence and the document lifecycle, rather than merely polishing a phrase and declaring the problem solved.

13. The denominator vanishes

Failure: A percentage loses the population counted. This can survive ordinary spell-checking and even a fluent bilingual read because the missing distinction sits in the interaction between wording, protocol context, presentation and the reader’s likely decision. Ask what the affected person would do differently if they trusted this version rather than the original.

Repair: Always retain population, observation window and numerator/denominator context. Then update the termbase or quality checklist with a narrowly scoped rule; assign the source owner any scientific ambiguity. Record whether a released version needs withdrawal or correction. A robust fix controls both the sentence and the document lifecycle, rather than merely polishing a phrase and declaring the problem solved.

14. The schedule shifts

Failure: A week, visit window or follow-up interval changes in translation. This can survive ordinary spell-checking and even a fluent bilingual read because the missing distinction sits in the interaction between wording, protocol context, presentation and the reader’s likely decision. Ask what the affected person would do differently if they trusted this version rather than the original.

Repair: Cross-compare with schedule of activities and time units. Then update the termbase or quality checklist with a narrowly scoped rule; assign the source owner any scientific ambiguity. Record whether a released version needs withdrawal or correction. A robust fix controls both the sentence and the document lifecycle, rather than merely polishing a phrase and declaring the problem solved.

15. A number changes meaning

Failure: Decimal mark, range, dose unit or inequality is transcribed wrongly. This can survive ordinary spell-checking and even a fluent bilingual read because the missing distinction sits in the interaction between wording, protocol context, presentation and the reader’s likely decision. Ask what the affected person would do differently if they trusted this version rather than the original.

Repair: Use a numerical QA pass independent of prose review. Then update the termbase or quality checklist with a narrowly scoped rule; assign the source owner any scientific ambiguity. Record whether a released version needs withdrawal or correction. A robust fix controls both the sentence and the document lifecycle, rather than merely polishing a phrase and declaring the problem solved.

16. A placeholder is treated as a contact

Failure: A fictional number or email is published as an operational support channel. This can survive ordinary spell-checking and even a fluent bilingual read because the missing distinction sits in the interaction between wording, protocol context, presentation and the reader’s likely decision. Ask what the affected person would do differently if they trusted this version rather than the original.

Repair: Require approved real contact channels before participant distribution. Then update the termbase or quality checklist with a narrowly scoped rule; assign the source owner any scientific ambiguity. Record whether a released version needs withdrawal or correction. A robust fix controls both the sentence and the document lifecycle, rather than merely polishing a phrase and declaring the problem solved.

17. The wrong document version ships

Failure: A superseded consent form remains on a recruitment site. This can survive ordinary spell-checking and even a fluent bilingual read because the missing distinction sits in the interaction between wording, protocol context, presentation and the reader’s likely decision. Ask what the affected person would do differently if they trusted this version rather than the original.

Repair: Reconcile document ID, source and translation versions, approval and effective dates. Then update the termbase or quality checklist with a narrowly scoped rule; assign the source owner any scientific ambiguity. Record whether a released version needs withdrawal or correction. A robust fix controls both the sentence and the document lifecycle, rather than merely polishing a phrase and declaring the problem solved.

18. Acronyms multiply

Failure: Different target terms are invented for one research concept. This can survive ordinary spell-checking and even a fluent bilingual read because the missing distinction sits in the interaction between wording, protocol context, presentation and the reader’s likely decision. Ask what the affected person would do differently if they trusted this version rather than the original.

Repair: Keep a glossary with allowed expansions and context. Then update the termbase or quality checklist with a narrowly scoped rule; assign the source owner any scientific ambiguity. Record whether a released version needs withdrawal or correction. A robust fix controls both the sentence and the document lifecycle, rather than merely polishing a phrase and declaring the problem solved.

19. A glossary becomes a protocol

Failure: A helpful definition adds steps or requirements not in the source. This can survive ordinary spell-checking and even a fluent bilingual read because the missing distinction sits in the interaction between wording, protocol context, presentation and the reader’s likely decision. Ask what the affected person would do differently if they trusted this version rather than the original.

Repair: Keep explanatory aids subordinate to approved research documents. Then update the termbase or quality checklist with a narrowly scoped rule; assign the source owner any scientific ambiguity. Record whether a released version needs withdrawal or correction. A robust fix controls both the sentence and the document lifecycle, rather than merely polishing a phrase and declaring the problem solved.

20. Interpreting is omitted

Failure: A translated leaflet is assumed to replace a live discussion. This can survive ordinary spell-checking and even a fluent bilingual read because the missing distinction sits in the interaction between wording, protocol context, presentation and the reader’s likely decision. Ask what the affected person would do differently if they trusted this version rather than the original.

Repair: Plan qualified communication support according to actual participant needs and governing rules. Then update the termbase or quality checklist with a narrowly scoped rule; assign the source owner any scientific ambiguity. Record whether a released version needs withdrawal or correction. A robust fix controls both the sentence and the document lifecycle, rather than merely polishing a phrase and declaring the problem solved.

21. The survey scale is reversed

Failure: Strongly agree and strongly disagree swap order or score values. This can survive ordinary spell-checking and even a fluent bilingual read because the missing distinction sits in the interaction between wording, protocol context, presentation and the reader’s likely decision. Ask what the affected person would do differently if they trusted this version rather than the original.

Repair: Test response anchors, code mappings and skip logic. Then update the termbase or quality checklist with a narrowly scoped rule; assign the source owner any scientific ambiguity. Record whether a released version needs withdrawal or correction. A robust fix controls both the sentence and the document lifecycle, rather than merely polishing a phrase and declaring the problem solved.

22. Eligibility is guessed

Failure: A translator resolves an ambiguous exclusion criterion alone. This can survive ordinary spell-checking and even a fluent bilingual read because the missing distinction sits in the interaction between wording, protocol context, presentation and the reader’s likely decision. Ask what the affected person would do differently if they trusted this version rather than the original.

Repair: Escalate to the authorised protocol owner for a documented clarification. Then update the termbase or quality checklist with a narrowly scoped rule; assign the source owner any scientific ambiguity. Record whether a released version needs withdrawal or correction. A robust fix controls both the sentence and the document lifecycle, rather than merely polishing a phrase and declaring the problem solved.

23. Registry status is stale

Failure: A recruiting page remains live after suspension or completion. This can survive ordinary spell-checking and even a fluent bilingual read because the missing distinction sits in the interaction between wording, protocol context, presentation and the reader’s likely decision. Ask what the affected person would do differently if they trusted this version rather than the original.

Repair: Recheck the source record and publication lifecycle. Then update the termbase or quality checklist with a narrowly scoped rule; assign the source owner any scientific ambiguity. Record whether a released version needs withdrawal or correction. A robust fix controls both the sentence and the document lifecycle, rather than merely polishing a phrase and declaring the problem solved.

24. Statistics are promotional

Failure: A subgroup estimate becomes the headline treatment effect. This can survive ordinary spell-checking and even a fluent bilingual read because the missing distinction sits in the interaction between wording, protocol context, presentation and the reader’s likely decision. Ask what the affected person would do differently if they trusted this version rather than the original.

Repair: Keep predefined endpoint, analysis population and uncertainty intact. Then update the termbase or quality checklist with a narrowly scoped rule; assign the source owner any scientific ambiguity. Record whether a released version needs withdrawal or correction. A robust fix controls both the sentence and the document lifecycle, rather than merely polishing a phrase and declaring the problem solved.

25. Privacy terms are overstated

Failure: Coded or pseudonymised records are called completely anonymous. This can survive ordinary spell-checking and even a fluent bilingual read because the missing distinction sits in the interaction between wording, protocol context, presentation and the reader’s likely decision. Ask what the affected person would do differently if they trusted this version rather than the original.

Repair: Confirm whether re-identification remains technically possible under controlled access. Then update the termbase or quality checklist with a narrowly scoped rule; assign the source owner any scientific ambiguity. Record whether a released version needs withdrawal or correction. A robust fix controls both the sentence and the document lifecycle, rather than merely polishing a phrase and declaring the problem solved.

26. Electronic controls disappear

Failure: An eConsent translation breaks branching, focus order or signatures. This can survive ordinary spell-checking and even a fluent bilingual read because the missing distinction sits in the interaction between wording, protocol context, presentation and the reader’s likely decision. Ask what the affected person would do differently if they trusted this version rather than the original.

Repair: Test the actual interface and record its version before release. Then update the termbase or quality checklist with a narrowly scoped rule; assign the source owner any scientific ambiguity. Record whether a released version needs withdrawal or correction. A robust fix controls both the sentence and the document lifecycle, rather than merely polishing a phrase and declaring the problem solved.

27. Automated translation is treated as signed off

Failure: Raw machine output is posted because it looks fluent. This can survive ordinary spell-checking and even a fluent bilingual read because the missing distinction sits in the interaction between wording, protocol context, presentation and the reader’s likely decision. Ask what the affected person would do differently if they trusted this version rather than the original.

Repair: Route it through appropriate expert checks, risk assessment and approval. Then update the termbase or quality checklist with a narrowly scoped rule; assign the source owner any scientific ambiguity. Record whether a released version needs withdrawal or correction. A robust fix controls both the sentence and the document lifecycle, rather than merely polishing a phrase and declaring the problem solved.

28. A correction changes the past

Failure: An amended risk statement replaces an archived version without trace. This can survive ordinary spell-checking and even a fluent bilingual read because the missing distinction sits in the interaction between wording, protocol context, presentation and the reader’s likely decision. Ask what the affected person would do differently if they trusted this version rather than the original.

Repair: Retain time-stamped history and make current-versus-previous status clear. Then update the termbase or quality checklist with a narrowly scoped rule; assign the source owner any scientific ambiguity. Record whether a released version needs withdrawal or correction. A robust fix controls both the sentence and the document lifecycle, rather than merely polishing a phrase and declaring the problem solved.

Worked diagnostics: 190 fictional clinical research translation decisions

Each case below uses a fictional English source line, an illustrative mistaken inference, a safer meaning-preserving direction and a concrete verification test. These are not certified target-language translations or approved consent statements. They teach the questions a translator and reviewer must resolve before producing any language version. Never distribute a teaching example as a real clinical trial instruction.

Voluntary consent and participant choice

The core test of consent translation is whether a person can distinguish an invitation from an instruction and a choice from a penalty. These cases use fictional source excerpts. They illustrate linguistic diagnosis only; the approved text and actual consent process govern real recruitment.

Diagnostic 001: Joining the study is your choice.

Controlled source excerpt (fictional): “Joining the study is your choice.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “You must participate to receive help.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Participation is voluntary. The incorrect reading removes the participant’s decision and can create pressure unrelated to the research purpose. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Ask a reader whether saying no changes the meaning of ordinary care. Second lens — the evidence-status check: Identify whether the phrase reports a plan, a possibility, an observation, a provisional assessment, an approved decision or a final result. Preserve that evidence level. Do not use later information to make the source sound more certain. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 002: You may take as much time as you need to decide, within the stated enrolment …

Controlled source excerpt (fictional): “You may take as much time as you need to decide, within the stated enrolment window.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Please sign immediately.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: You may consider the information before deciding, subject to the stated recruitment window. The wrong version suppresses deliberation; the correct version keeps both freedom and the real time limit. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Compare the window with the current approved recruitment materials. Second lens — the role-and-authority check: Identify the actor named in the controlled record. A study nurse, participant, investigator, sponsor, ethics committee and data system do different things. A translation that transfers power to another actor can be wrong without altering any technical noun. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 003: You can ask questions before agreeing.

Controlled source excerpt (fictional): “You can ask questions before agreeing.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Questions will be answered after you join.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: You may ask questions before making a decision. The mistaken sequence turns discussion into a post-enrolment privilege rather than part of the consent process. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Test what the reader expects at the meeting. Second lens — the scope-and-exception check: Underline every qualifier: if, unless, only, at least, no later than, optional, possible, expected, in the named group or under the stated circumstances. Test whether the target sentence covers exactly the same cases. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 004: Refusing participation will not affect your usual care.

Controlled source excerpt (fictional): “Refusing participation will not affect your usual care.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Refusing means we cannot continue your care.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Your usual care is not affected by your decision not to join. A translated threat can induce consent and make the nature of ordinary care unclear. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Confirm the exact approved assurance and its scope. Second lens — the source-version check: Record the document identifier and revision. Match the line to the currently authorised study document; do not repair the scientific source by intuition. If the source conflicts with another controlled file, open a clarification ticket. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 005: You may stop taking part in study activities.

Controlled source excerpt (fictional): “You may stop taking part in study activities.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “You may discontinue all clinical care.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: You can stop participating in the research activities as explained. Conflating research withdrawal with clinical follow-up may affect patient safety and trust. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Verify that no clinically necessary continuity instructions were added or erased. Second lens — the numeric-and-structured-data check: Compare figures, units, groups, time windows, codes, ordered answers, signatures and references separately from prose. The correct target-language sentence can still be dangerous if an adjacent field contains a copied numerical error. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 006: We may ask your permission to contact you about future studies.

Controlled source excerpt (fictional): “We may ask your permission to contact you about future studies.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “You agree to all future studies.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: We may ask separately whether we can contact you about future research. Permission for contact is narrower than permission to enrol or reuse data. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Review each checkbox and optional choice separately. Second lens — the comprehension check: Ask the intended audience to explain what they understand without prompting. They should be able to describe the relevant choice or evidence in ordinary words. Misunderstanding is diagnostic feedback, not a reason to simplify away a right or a risk. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 007: Your decision will not affect your relationship with the research team.

Controlled source excerpt (fictional): “Your decision will not affect your relationship with the research team.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The team will never contact you again if you decline.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Declining will not change how the research team treats you in the stated context. A social reassurance is not a ban on future permissible communication. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Keep the source’s exact relationship claim and avoid promises. Second lens — the cross-document check: Look for the same concept in the information sheet, protocol, schedule, form, registry and results text. If one version describes a different action or population, establish whether the source documents legitimately differ or whether the translation has drifted. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 008: You may discuss this document with someone you trust.

Controlled source excerpt (fictional): “You may discuss this document with someone you trust.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “A family member must authorise participation.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: You may consult someone you trust before deciding. Advice and formal legal representation are different roles. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check that no family-consent requirement was invented. Second lens — the release-state check: A finished translation is not automatically approved for use. Verify scientific review, language review, required approvals, version status and distribution scope, following the study’s governance and applicable requirements. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 009: We will explain any new information relevant to your willingness to continue.

Controlled source excerpt (fictional): “We will explain any new information relevant to your willingness to continue.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “We guarantee the study will never change.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Relevant new information will be communicated as described. Research knowledge evolves and the participant may need to revisit a decision. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Compare with amendment and reconsent provisions. Second lens — the reader-decision check: Picture a person who never sees the original. Would this wording cause them to consent, decline, report, stop, sign, attend, share information or expect a benefit on a different basis? If so, the divergence is substantive even when grammar is flawless. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 010: Signing indicates that the consent process was documented.

Controlled source excerpt (fictional): “Signing indicates that the consent process was documented.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Signing proves you understand every medical detail.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Signing documents the specified consent step; it is not a substitute for a meaningful discussion. A signature alone is not an adequate comprehension test. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Ask whether the target makes a stronger claim than the source. Second lens — the evidence-status check: Identify whether the phrase reports a plan, a possibility, an observation, a provisional assessment, an approved decision or a final result. Preserve that evidence level. Do not use later information to make the source sound more certain. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Potential benefits, risks and uncertainty

Words that make a trial sound promising can be inaccurate without being overtly promotional. Preserve whether the source describes a possible benefit, a known risk, an unmeasured effect or an unknown risk. Do not translate probabilistic language as personal medical advice.

Diagnostic 011: You may not benefit personally from this study.

Controlled source excerpt (fictional): “You may not benefit personally from this study.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The study will improve your health.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: You may receive no direct personal benefit. The incorrect text sells expected improvement that the research does not guarantee. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Compare every benefit verb with the consent source. Second lens — the role-and-authority check: Identify the actor named in the controlled record. A study nurse, participant, investigator, sponsor, ethics committee and data system do different things. A translation that transfers power to another actor can be wrong without altering any technical noun. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 012: Some risks are not yet known.

Controlled source excerpt (fictional): “Some risks are not yet known.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “All possible risks have been identified.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Some risks remain unknown. Uncertainty is a research fact, not an editorial imperfection to eliminate. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check that the unknown-risk sentence remains visible and unqualified. Second lens — the scope-and-exception check: Underline every qualifier: if, unless, only, at least, no later than, optional, possible, expected, in the named group or under the stated circumstances. Test whether the target sentence covers exactly the same cases. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 013: The procedure may cause temporary discomfort.

Controlled source excerpt (fictional): “The procedure may cause temporary discomfort.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The procedure is painless.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The procedure can cause temporary discomfort. A reassurance about absence of pain removes a foreseeable burden. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Read the procedure description beside the risk paragraph. Second lens — the source-version check: Record the document identifier and revision. Match the line to the currently authorised study document; do not repair the scientific source by intuition. If the source conflicts with another controlled file, open a clarification ticket. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 014: A serious complication is possible but uncommon, according to the source info…

Controlled source excerpt (fictional): “A serious complication is possible but uncommon, according to the source information.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The complication cannot happen.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: A serious complication is possible; retain the source’s supported frequency wording. False impossibility and unsourced precision both distort choice. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check whether the cited frequency has a study-defined denominator. Second lens — the comprehension check: Ask the intended audience to explain what they understand without prompting. They should be able to describe the relevant choice or evidence in ordinary words. Misunderstanding is diagnostic feedback, not a reason to simplify away a right or a risk. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 015: We do not know whether the intervention is more effective than the comparator.

Controlled source excerpt (fictional): “We do not know whether the intervention is more effective than the comparator.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The intervention works better than the comparator.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: It is not yet known whether it is more effective than the comparator. A clinical hypothesis is not an established outcome. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Ask whether the target describes an unanswered question. Second lens — the escalation check: Some ambiguity is not translatable until the source owner answers it. Record the question, candidate meanings, clinical impact, responsible reviewer and authorised resolution. A silent guess is not quality assurance. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 016: The information available so far is limited.

Controlled source excerpt (fictional): “The information available so far is limited.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The intervention has been thoroughly proven safe.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Current information is limited in the stated respects. Safety cannot be guaranteed from limited evidence. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Preserve the source’s actual evidence scope. Second lens — the cross-document check: Look for the same concept in the information sheet, protocol, schedule, form, registry and results text. If one version describes a different action or population, establish whether the source documents legitimately differ or whether the translation has drifted. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 017: There is a chance you will experience a side effect.

Controlled source excerpt (fictional): “There is a chance you will experience a side effect.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “You will experience every listed side effect.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: You may experience one or more listed side effects. Possibility must not become inevitability; both can deter appropriate voluntary consideration. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Read the source’s level of certainty. Second lens — the release-state check: A finished translation is not automatically approved for use. Verify scientific review, language review, required approvals, version status and distribution scope, following the study’s governance and applicable requirements. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 018: The study seeks to learn about future patients, not to promise treatment to v…

Controlled source excerpt (fictional): “The study seeks to learn about future patients, not to promise treatment to volunteers.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Every volunteer receives treatment benefits.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The research aims to produce knowledge; individual benefit is not guaranteed. This preserves the public aim and the limits of individual expectation. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check the purpose paragraph and benefit section together. Second lens — the reader-decision check: Picture a person who never sees the original. Would this wording cause them to consent, decline, report, stop, sign, attend, share information or expect a benefit on a different basis? If so, the divergence is substantive even when grammar is flawless. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 019: The study team will monitor for specified safety signals.

Controlled source excerpt (fictional): “The study team will monitor for specified safety signals.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The study team prevents all adverse events.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The team monitors for specified safety events as described. Observation is not prevention and must not be described as a shield against risk. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check monitoring schedule against claims. Second lens — the evidence-status check: Identify whether the phrase reports a plan, a possibility, an observation, a provisional assessment, an approved decision or a final result. Preserve that evidence level. Do not use later information to make the source sound more certain. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 020: If new risks become known, the team may update the information.

Controlled source excerpt (fictional): “If new risks become known, the team may update the information.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The original risk list is final.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Risk information may be updated when relevant new evidence emerges. The target must allow the source’s revision process. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Tie new information to the appropriate controlled document. Second lens — the role-and-authority check: Identify the actor named in the controlled record. A study nurse, participant, investigator, sponsor, ethics committee and data system do different things. A translation that transfers power to another actor can be wrong without altering any technical noun. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Randomisation, blinding and comparison groups

Random assignment answers a different question from whether an intervention works. Translate assignment methods, masks and comparators as described, not as universal assumptions about clinical trials.

Diagnostic 021: Participants will be assigned to a group by chance.

Controlled source excerpt (fictional): “Participants will be assigned to a group by chance.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The doctor will choose the best group for you.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: A chance process assigns the group. A translator must not turn experimental allocation into personalised clinical selection. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Confirm the protocol’s allocation mechanism. Second lens — the source-version check: Record the document identifier and revision. Match the line to the currently authorised study document; do not repair the scientific source by intuition. If the source conflicts with another controlled file, open a clarification ticket. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 022: Each group has an equal chance of assignment under this trial design.

Controlled source excerpt (fictional): “Each group has an equal chance of assignment under this trial design.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Every participant gets both groups.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The described design gives each eligible participant an equal chance of either group. Probability of assignment is not receipt of multiple interventions. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check the actual allocation ratio and group count. Second lens — the numeric-and-structured-data check: Compare figures, units, groups, time windows, codes, ordered answers, signatures and references separately from prose. The correct target-language sentence can still be dangerous if an adjacent field contains a copied numerical error. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 023: Participants will not know which group they are in.

Controlled source excerpt (fictional): “Participants will not know which group they are in.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “No one in the study knows the assignment.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Participants are masked; do not extend masking to other roles unless stated. Masking varies across roles and procedures. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check the source’s masking description. Second lens — the comprehension check: Ask the intended audience to explain what they understand without prompting. They should be able to describe the relevant choice or evidence in ordinary words. Misunderstanding is diagnostic feedback, not a reason to simplify away a right or a risk. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 024: The assessor is blinded to the assignment.

Controlled source excerpt (fictional): “The assessor is blinded to the assignment.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The participant is blinded to the diagnosis.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The outcome assessor does not know the assigned group. Assessor masking is narrower than participant masking or diagnostic uncertainty. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Identify who is masked and from which fact. Second lens — the escalation check: Some ambiguity is not translatable until the source owner answers it. Record the question, candidate meanings, clinical impact, responsible reviewer and authorised resolution. A silent guess is not quality assurance. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 025: A placebo contains no active study medicine.

Controlled source excerpt (fictional): “A placebo contains no active study medicine.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “A placebo means no other care is given.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The placebo is described as lacking the active study intervention. Placebo description does not tell the whole care plan. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Compare standard-care and rescue-treatment provisions. Second lens — the cross-document check: Look for the same concept in the information sheet, protocol, schedule, form, registry and results text. If one version describes a different action or population, establish whether the source documents legitimately differ or whether the translation has drifted. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 026: The comparator group receives standard treatment as described.

Controlled source excerpt (fictional): “The comparator group receives standard treatment as described.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The control group receives nothing.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The comparison group receives the stated standard treatment. The unsafe version denies care that the design actually includes. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Cross-check group descriptions against the schedule. Second lens — the release-state check: A finished translation is not automatically approved for use. Verify scientific review, language review, required approvals, version status and distribution scope, following the study’s governance and applicable requirements. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 027: Open-label means the relevant allocation is not masked.

Controlled source excerpt (fictional): “Open-label means the relevant allocation is not masked.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Open-label means results are public immediately.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The stated treatment assignment is known to the roles described. Disclosure of assignment is not publication of study outcomes. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Test what a target-language reader thinks open-label permits. Second lens — the reader-decision check: Picture a person who never sees the original. Would this wording cause them to consent, decline, report, stop, sign, attend, share information or expect a benefit on a different basis? If so, the divergence is substantive even when grammar is flawless. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 028: Treatment assignments may be unblinded in specified circumstances.

Controlled source excerpt (fictional): “Treatment assignments may be unblinded in specified circumstances.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Anyone can reveal assignments at any time.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Unmasking follows the source’s specified conditions and authority. A conditional safety procedure must not become unlimited permission. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check who can authorise and document unblinding. Second lens — the role-and-authority check: Identify the actor named in the controlled record. A study nurse, participant, investigator, sponsor, ethics committee and data system do different things. A translation that transfers power to another actor can be wrong without altering any technical noun. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 029: The trial includes an active comparator.

Controlled source excerpt (fictional): “The trial includes an active comparator.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The comparator is a neutral placebo.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The comparison is another active intervention as stated. Replacing active control with placebo changes the ethics and scientific question. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Inspect the intervention arm table. Second lens — the scope-and-exception check: Underline every qualifier: if, unless, only, at least, no later than, optional, possible, expected, in the named group or under the stated circumstances. Test whether the target sentence covers exactly the same cases. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 030: Allocation is concealed until the prescribed point.

Controlled source excerpt (fictional): “Allocation is concealed until the prescribed point.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The assigned treatment never exists in records.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Assignment information is protected from specified roles until the defined stage. Concealment and record retention are not opposites. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check sequence and operational role access. Second lens — the source-version check: Record the document identifier and revision. Match the line to the currently authorised study document; do not repair the scientific source by intuition. If the source conflicts with another controlled file, open a clarification ticket. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Eligibility, screening and recruitment

Clinical trial language often compresses statuses into everyday words such as accepted, selected or confirmed. A reader needs to know the difference between expressing interest, being screened, meeting criteria, enrolling and entering an analysis.

Diagnostic 031: Screening determines whether the study criteria are met.

Controlled source excerpt (fictional): “Screening determines whether the study criteria are met.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The appointment guarantees a place in the study.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Screening assesses whether a person meets the protocol’s eligibility requirements. An assessment is not a promise of enrolment. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Trace the status labels through invitations and screening forms. Second lens — the numeric-and-structured-data check: Compare figures, units, groups, time windows, codes, ordered answers, signatures and references separately from prose. The correct target-language sentence can still be dangerous if an adjacent field contains a copied numerical error. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 032: You may be excluded if a listed criterion applies.

Controlled source excerpt (fictional): “You may be excluded if a listed criterion applies.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “You have been diagnosed with the excluded condition.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: A listed circumstance can make a person ineligible, subject to the study’s assessment. An exclusion rule is not a clinical diagnosis. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check the source’s conditional wording. Second lens — the comprehension check: Ask the intended audience to explain what they understand without prompting. They should be able to describe the relevant choice or evidence in ordinary words. Misunderstanding is diagnostic feedback, not a reason to simplify away a right or a risk. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 033: Eligibility is confirmed only after required tests are reviewed.

Controlled source excerpt (fictional): “Eligibility is confirmed only after required tests are reviewed.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “You are eligible as soon as tests are taken.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Eligibility is determined after the specified results are assessed. The wrong wording removes an important gate. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Compare with the screening workflow. Second lens — the escalation check: Some ambiguity is not translatable until the source owner answers it. Record the question, candidate meanings, clinical impact, responsible reviewer and authorised resolution. A silent guess is not quality assurance. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 034: Recruitment is open at the time of this notice.

Controlled source excerpt (fictional): “Recruitment is open at the time of this notice.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Recruitment will always remain open.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The trial is recruiting as of the specified dated notice. A time-dependent status should not become a timeless promise. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Review live recruitment records before publication. Second lens — the cross-document check: Look for the same concept in the information sheet, protocol, schedule, form, registry and results text. If one version describes a different action or population, establish whether the source documents legitimately differ or whether the translation has drifted. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 035: The study is not recruiting new participants.

Controlled source excerpt (fictional): “The study is not recruiting new participants.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “No participants are involved in the study.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: New enrolment is closed, without claiming that existing follow-up has ended. Recruitment status is not trial completion. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check participant follow-up and registry status separately. Second lens — the reader-decision check: Picture a person who never sees the original. Would this wording cause them to consent, decline, report, stop, sign, attend, share information or expect a benefit on a different basis? If so, the divergence is substantive even when grammar is flawless. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 036: An exclusion criterion concerns current medication use.

Controlled source excerpt (fictional): “An exclusion criterion concerns current medication use.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The medication is medically forbidden for everyone.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The criterion determines study eligibility, not general clinical safety advice. Research eligibility must not pose as a universal prescription. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Consult the protocol term and counsel only through authorised clinicians. Second lens — the evidence-status check: Identify whether the phrase reports a plan, a possibility, an observation, a provisional assessment, an approved decision or a final result. Preserve that evidence level. Do not use later information to make the source sound more certain. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 037: A pre-screening call collects preliminary information.

Controlled source excerpt (fictional): “A pre-screening call collects preliminary information.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The call provides a medical diagnosis.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The call collects information for preliminary suitability assessment. Early recruitment cannot be reframed as medical evaluation. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Test readers’ expectations about what the call provides. Second lens — the role-and-authority check: Identify the actor named in the controlled record. A study nurse, participant, investigator, sponsor, ethics committee and data system do different things. A translation that transfers power to another actor can be wrong without altering any technical noun. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 038: Participants must meet all inclusion criteria and none of the exclusion crite…

Controlled source excerpt (fictional): “Participants must meet all inclusion criteria and none of the exclusion criteria.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “One inclusion criterion is enough.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The stated combination of inclusion and exclusion criteria governs eligibility. Logical operators can change the trial population. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Compare AND, OR and NOT against source conditions. Second lens — the scope-and-exception check: Underline every qualifier: if, unless, only, at least, no later than, optional, possible, expected, in the named group or under the stated circumstances. Test whether the target sentence covers exactly the same cases. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 039: An age range uses the stated limits inclusive or exclusive as defined.

Controlled source excerpt (fictional): “An age range uses the stated limits inclusive or exclusive as defined.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The study accepts all ages near the listed range.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Keep both age thresholds and boundary rules exactly as specified. Changing one boundary alters who may enrol. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check symbols, birth-date rules and age-at-event. Second lens — the source-version check: Record the document identifier and revision. Match the line to the currently authorised study document; do not repair the scientific source by intuition. If the source conflicts with another controlled file, open a clarification ticket. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 040: The investigator will confirm eligibility.

Controlled source excerpt (fictional): “The investigator will confirm eligibility.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The translator will decide whether you qualify.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Eligibility is assessed by the authorised study personnel. Translation communicates criteria but does not determine participant status. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Verify decision authority in the current protocol. Second lens — the numeric-and-structured-data check: Compare figures, units, groups, time windows, codes, ordered answers, signatures and references separately from prose. The correct target-language sentence can still be dangerous if an adjacent field contains a copied numerical error. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Study visits, procedures and participant burden

Schedules are full of hidden semantics: before versus after, optional versus required, window versus fixed date, and research-only procedure versus ordinary follow-up. Translate the table and the prose against each other.

Diagnostic 041: Visit 2 is planned for week four with a permitted window.

Controlled source excerpt (fictional): “Visit 2 is planned for week four with a permitted window.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Visit 2 must occur exactly on day 28.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Visit 2 is planned for the defined week-four window. A flexible protocol window must not become a new fixed deadline. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Compare schedule-of-activities notes. Second lens — the comprehension check: Ask the intended audience to explain what they understand without prompting. They should be able to describe the relevant choice or evidence in ordinary words. Misunderstanding is diagnostic feedback, not a reason to simplify away a right or a risk. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 042: A blood sample may be collected at the indicated visit.

Controlled source excerpt (fictional): “A blood sample may be collected at the indicated visit.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “A blood sample will be collected at every visit.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The sample collection is limited to the specified visit and conditional status. Extra perceived procedures change participant burden. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check table symbols and footnotes. Second lens — the cross-document check: Look for the same concept in the information sheet, protocol, schedule, form, registry and results text. If one version describes a different action or population, establish whether the source documents legitimately differ or whether the translation has drifted. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 043: The questionnaire is optional.

Controlled source excerpt (fictional): “The questionnaire is optional.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The questionnaire is compulsory for all participants.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The questionnaire may be completed voluntarily, as described. Optional participation cannot be erased by simplifying the instructions. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Test actual response-form settings. Second lens — the release-state check: A finished translation is not automatically approved for use. Verify scientific review, language review, required approvals, version status and distribution scope, following the study’s governance and applicable requirements. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 044: The follow-up call takes approximately the stated duration.

Controlled source excerpt (fictional): “The follow-up call takes approximately the stated duration.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The follow-up call always lasts that exact number of minutes.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The call is expected to take about the specified time. An estimate must retain its qualifier. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Compare with participant information and scheduling script. Second lens — the reader-decision check: Picture a person who never sees the original. Would this wording cause them to consent, decline, report, stop, sign, attend, share information or expect a benefit on a different basis? If so, the divergence is substantive even when grammar is flawless. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 045: A visit may be conducted remotely where approved.

Controlled source excerpt (fictional): “A visit may be conducted remotely where approved.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “All visits are virtual.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The specified visit may use an approved remote method when conditions permit. A conditional delivery option is not a global default. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check protocol amendments and site capabilities. Second lens — the evidence-status check: Identify whether the phrase reports a plan, a possibility, an observation, a provisional assessment, an approved decision or a final result. Preserve that evidence level. Do not use later information to make the source sound more certain. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 046: The participant may need to fast before a specified test.

Controlled source excerpt (fictional): “The participant may need to fast before a specified test.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Participants must not eat during the entire study.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Any fasting requirement applies only as stated for the named procedure. An overbroad instruction can affect wellbeing and compliance. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Escalate practical instructions to authorised clinical personnel. Second lens — the role-and-authority check: Identify the actor named in the controlled record. A study nurse, participant, investigator, sponsor, ethics committee and data system do different things. A translation that transfers power to another actor can be wrong without altering any technical noun. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 047: The imaging assessment is part of the research schedule.

Controlled source excerpt (fictional): “The imaging assessment is part of the research schedule.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The scan is a proven treatment.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The assessment is a study procedure, not itself a therapy. Misdescribing assessment as treatment changes expectations. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check imaging purpose and consent burden. Second lens — the scope-and-exception check: Underline every qualifier: if, unless, only, at least, no later than, optional, possible, expected, in the named group or under the stated circumstances. Test whether the target sentence covers exactly the same cases. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 048: A home diary records daily symptoms.

Controlled source excerpt (fictional): “A home diary records daily symptoms.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The diary replaces the investigator’s assessment.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The diary is a participant-reported record as specified. Self-report and clinician assessment are distinct evidence streams. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check diary items and reporting destination. Second lens — the source-version check: Record the document identifier and revision. Match the line to the currently authorised study document; do not repair the scientific source by intuition. If the source conflicts with another controlled file, open a clarification ticket. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 049: An unscheduled visit may be requested for safety reasons.

Controlled source excerpt (fictional): “An unscheduled visit may be requested for safety reasons.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Any visit can be skipped without discussion.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The protocol allows an additional safety visit in specified circumstances. The unsafe reading alters escalation and visit meaning. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Preserve authorisation and triggers. Second lens — the comprehension check: Ask the intended audience to explain what they understand without prompting. They should be able to describe the relevant choice or evidence in ordinary words. Misunderstanding is diagnostic feedback, not a reason to simplify away a right or a risk. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 050: Samples may be stored for the stated period.

Controlled source excerpt (fictional): “Samples may be stored for the stated period.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Samples will be kept forever.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Storage duration and conditions are exactly those described in the approved material. Duration can matter to participant choice. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Verify any later optional consent separately. Second lens — the escalation check: Some ambiguity is not translatable until the source owner answers it. Record the question, candidate meanings, clinical impact, responsible reviewer and authorised resolution. A silent guess is not quality assurance. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Interventions, dosage descriptions and concomitant treatment

Translator-facing instruction is not patient dosing advice. Preserve numbers, formulation labels, group assignment and investigator responsibility; never infer or improvise instructions from a fictional example.

Diagnostic 051: The protocol names a once-daily investigational product schedule.

Controlled source excerpt (fictional): “The protocol names a once-daily investigational product schedule.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The participant can take an extra dose if desired.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The study protocol specifies the dosing schedule without discretionary extra doses. Invented options can create a safety hazard. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Compare the exact controlled instructions under qualified review. Second lens — the cross-document check: Look for the same concept in the information sheet, protocol, schedule, form, registry and results text. If one version describes a different action or population, establish whether the source documents legitimately differ or whether the translation has drifted. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 052: A dose is expressed in milligrams, not millilitres.

Controlled source excerpt (fictional): “A dose is expressed in milligrams, not millilitres.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The same number of millilitres is interchangeable.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Retain the unit and formulation relationship exactly as written. Mass and volume cannot be swapped by changing a label. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Have a clinical reviewer verify numeric fidelity. Second lens — the release-state check: A finished translation is not automatically approved for use. Verify scientific review, language review, required approvals, version status and distribution scope, following the study’s governance and applicable requirements. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 053: The intervention is administered by trained staff.

Controlled source excerpt (fictional): “The intervention is administered by trained staff.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Participants administer it at home by default.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The stated qualified staff administer the intervention. Changing the actor changes responsibility and risk controls. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Match administration setting and role. Second lens — the reader-decision check: Picture a person who never sees the original. Would this wording cause them to consent, decline, report, stop, sign, attend, share information or expect a benefit on a different basis? If so, the divergence is substantive even when grammar is flawless. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 054: The comparator is a protocol-defined active treatment.

Controlled source excerpt (fictional): “The comparator is a protocol-defined active treatment.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The comparator is simply a sugar pill.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The comparator is the specified active therapy. The experimental comparison changes when the comparator is misidentified. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Inspect group labels. Second lens — the evidence-status check: Identify whether the phrase reports a plan, a possibility, an observation, a provisional assessment, an approved decision or a final result. Preserve that evidence level. Do not use later information to make the source sound more certain. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 055: Concomitant medications must be reported as specified.

Controlled source excerpt (fictional): “Concomitant medications must be reported as specified.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “All other medicines are automatically banned.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Participants report other medication use according to the study’s criteria. Reporting is not a universal prohibition. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Leave medication decisions to treating professionals. Second lens — the role-and-authority check: Identify the actor named in the controlled record. A study nurse, participant, investigator, sponsor, ethics committee and data system do different things. A translation that transfers power to another actor can be wrong without altering any technical noun. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 056: A treatment cycle consists of a defined schedule and rest interval.

Controlled source excerpt (fictional): “A treatment cycle consists of a defined schedule and rest interval.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The cycle is any convenient number of days.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Keep cycle boundaries and rest interval as stated. A shortened cycle changes exposure and visit timing. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Validate cycle ID and dates. Second lens — the source-version check: Record the document identifier and revision. Match the line to the currently authorised study document; do not repair the scientific source by intuition. If the source conflicts with another controlled file, open a clarification ticket. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 057: The source states an infusion duration with a range.

Controlled source excerpt (fictional): “The source states an infusion duration with a range.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The infusion is always completed immediately.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Preserve the specified duration range and qualifying conditions. Removing duration changes the apparent procedure. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Send any clinical ambiguity for specialist resolution. Second lens — the numeric-and-structured-data check: Compare figures, units, groups, time windows, codes, ordered answers, signatures and references separately from prose. The correct target-language sentence can still be dangerous if an adjacent field contains a copied numerical error. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 058: Dose reduction may occur under protocol criteria.

Controlled source excerpt (fictional): “Dose reduction may occur under protocol criteria.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Participants should reduce their own dose when worried.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Any modification is governed by the authorised protocol and personnel. A conditional research rule must not become self-treatment advice. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Confirm actor, trigger and procedure. Second lens — the comprehension check: Ask the intended audience to explain what they understand without prompting. They should be able to describe the relevant choice or evidence in ordinary words. Misunderstanding is diagnostic feedback, not a reason to simplify away a right or a risk. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 059: Rescue medication is permitted under specified circumstances.

Controlled source excerpt (fictional): “Rescue medication is permitted under specified circumstances.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Rescue medication replaces all assigned study care.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Its permitted use is limited to the conditions described. The term has a protocol function rather than an everyday promise of rescue. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check comparator and concomitant-use clauses. Second lens — the escalation check: Some ambiguity is not translatable until the source owner answers it. Record the question, candidate meanings, clinical impact, responsible reviewer and authorised resolution. A silent guess is not quality assurance. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 060: The investigational product may be discontinued for stated reasons.

Controlled source excerpt (fictional): “The investigational product may be discontinued for stated reasons.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “All trial participation ends automatically when product use stops.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Stopping study product and continuing follow-up are distinct when the source says so. The wrong version can obscure continuing assessments. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check discontinuation versus withdrawal terminology. Second lens — the cross-document check: Look for the same concept in the information sheet, protocol, schedule, form, registry and results text. If one version describes a different action or population, establish whether the source documents legitimately differ or whether the translation has drifted. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Adverse events, seriousness and attribution

An adverse event is an occurrence, while seriousness, severity, expectedness and relatedness are separate properties. Translate the labelled category actually used by the source; do not impose clinical conclusions.

Diagnostic 061: An adverse event was reported during follow-up.

Controlled source excerpt (fictional): “An adverse event was reported during follow-up.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The treatment caused an adverse reaction.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: An event was reported during follow-up without a claim of causality. Sequence does not establish cause. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Review whether attribution was assessed. Second lens — the release-state check: A finished translation is not automatically approved for use. Verify scientific review, language review, required approvals, version status and distribution scope, following the study’s governance and applicable requirements. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 062: The event was graded as severe under the specified scale.

Controlled source excerpt (fictional): “The event was graded as severe under the specified scale.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The event automatically meets serious adverse event criteria.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The event’s intensity was graded as severe on the named scale. Severity and regulatory seriousness use different criteria. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Preserve the scale name and classification. Second lens — the reader-decision check: Picture a person who never sees the original. Would this wording cause them to consent, decline, report, stop, sign, attend, share information or expect a benefit on a different basis? If so, the divergence is substantive even when grammar is flawless. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 063: The occurrence met the seriousness definition in the study records.

Controlled source excerpt (fictional): “The occurrence met the seriousness definition in the study records.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The symptom was necessarily the most intense possible.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The record classifies the event as serious under applicable criteria. Serious is not a synonym for maximum symptom intensity. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Verify the reason for classification. Second lens — the role-and-authority check: Identify the actor named in the controlled record. A study nurse, participant, investigator, sponsor, ethics committee and data system do different things. A translation that transfers power to another actor can be wrong without altering any technical noun. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 064: Relatedness was assessed as unlikely.

Controlled source excerpt (fictional): “Relatedness was assessed as unlikely.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The study proves the intervention could not have contributed.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The recorded assessment considered a relationship unlikely. A likelihood judgement is not absolute exclusion. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Maintain the assessor and judgement date. Second lens — the scope-and-exception check: Underline every qualifier: if, unless, only, at least, no later than, optional, possible, expected, in the named group or under the stated circumstances. Test whether the target sentence covers exactly the same cases. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 065: The adverse event was unexpected relative to reference information.

Controlled source excerpt (fictional): “The adverse event was unexpected relative to reference information.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “No one expected any adverse event in a trial.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Unexpectedness is defined against the stated reference safety information. Everyday surprise is not necessarily a regulatory classification. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check the reference version. Second lens — the source-version check: Record the document identifier and revision. Match the line to the currently authorised study document; do not repair the scientific source by intuition. If the source conflicts with another controlled file, open a clarification ticket. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 066: The incident was first reported on the stated date.

Controlled source excerpt (fictional): “The incident was first reported on the stated date.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The event first occurred on the reporting date.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The report date is separate from the actual event-onset date. Timing changes the interpretation of event history. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Review date fields individually. Second lens — the numeric-and-structured-data check: Compare figures, units, groups, time windows, codes, ordered answers, signatures and references separately from prose. The correct target-language sentence can still be dangerous if an adjacent field contains a copied numerical error. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 067: The event resolved with reported follow-up.

Controlled source excerpt (fictional): “The event resolved with reported follow-up.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “No harm ever occurred.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The event subsequently resolved as recorded. Resolution does not erase the earlier occurrence. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Preserve start, outcome and resolution fields. Second lens — the comprehension check: Ask the intended audience to explain what they understand without prompting. They should be able to describe the relevant choice or evidence in ordinary words. Misunderstanding is diagnostic feedback, not a reason to simplify away a right or a risk. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 068: An investigator assessed the relationship.

Controlled source excerpt (fictional): “An investigator assessed the relationship.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “A translator confirmed the diagnosis and cause.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The investigator made the recorded assessment. Authority to classify an event remains with qualified personnel. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check role attribution. Second lens — the escalation check: Some ambiguity is not translatable until the source owner answers it. Record the question, candidate meanings, clinical impact, responsible reviewer and authorised resolution. A silent guess is not quality assurance. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 069: An event led to study-drug discontinuation.

Controlled source excerpt (fictional): “An event led to study-drug discontinuation.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The participant withdrew permission for all further research contact.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Drug discontinuation occurred; any broader withdrawal must be separately evidenced. A product decision is not necessarily a consent decision. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Review discontinuation reasons. Second lens — the cross-document check: Look for the same concept in the information sheet, protocol, schedule, form, registry and results text. If one version describes a different action or population, establish whether the source documents legitimately differ or whether the translation has drifted. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 070: The report concerns a suspected unexpected serious adverse reaction.

Controlled source excerpt (fictional): “The report concerns a suspected unexpected serious adverse reaction.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Every serious report is proven caused by the product.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Retain suspected, unexpected, serious and adverse reaction as distinct qualifiers. Every qualifier contributes to the classification. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Compare with controlled pharmacovigilance terminology. Second lens — the reader-decision check: Picture a person who never sees the original. Would this wording cause them to consent, decline, report, stop, sign, attend, share information or expect a benefit on a different basis? If so, the divergence is substantive even when grammar is flawless. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Safety reporting, monitoring and urgent communications

A safety notice is actionable language. What happened, who knows, who must act, and when each report is due may be distributed across a form. These teaching examples are not live clinical safety instructions.

Diagnostic 071: Report the listed event promptly through the study contact stated in the appr…

Controlled source excerpt (fictional): “Report the listed event promptly through the study contact stated in the approved document.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Wait until the next routine visit for every event.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Follow the specified reporting instruction and contact channel. A shift in urgency can delay needed communication. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Validate real contact information separately from sample text. Second lens — the evidence-status check: Identify whether the phrase reports a plan, a possibility, an observation, a provisional assessment, an approved decision or a final result. Preserve that evidence level. Do not use later information to make the source sound more certain. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 072: The sponsor received the initial report at the stated time.

Controlled source excerpt (fictional): “The sponsor received the initial report at the stated time.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The regulator confirmed the report at that time.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The recorded event is sponsor receipt, not regulatory assessment. Different workflow timestamps represent distinct actions. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Inspect the event audit trail. Second lens — the role-and-authority check: Identify the actor named in the controlled record. A study nurse, participant, investigator, sponsor, ethics committee and data system do different things. A translation that transfers power to another actor can be wrong without altering any technical noun. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 073: A safety review is ongoing.

Controlled source excerpt (fictional): “A safety review is ongoing.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “A safety review has cleared the product.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The review is in progress; no conclusion is implied. Process status is not a judgement of safety. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Compare current status with decision record. Second lens — the scope-and-exception check: Underline every qualifier: if, unless, only, at least, no later than, optional, possible, expected, in the named group or under the stated circumstances. Test whether the target sentence covers exactly the same cases. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 074: The trial is temporarily paused to investigate a signal.

Controlled source excerpt (fictional): “The trial is temporarily paused to investigate a signal.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The study has been permanently cancelled for proven harm.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Recruitment or activities are paused as stated pending investigation. A pause has different scope from cancellation and causality. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check suspension notice and date. Second lens — the source-version check: Record the document identifier and revision. Match the line to the currently authorised study document; do not repair the scientific source by intuition. If the source conflicts with another controlled file, open a clarification ticket. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 075: An independent board recommended further monitoring.

Controlled source excerpt (fictional): “An independent board recommended further monitoring.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The board guaranteed no additional risks.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The board recommended the stated monitoring action. Recommendations are not safety guarantees. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Preserve who recommended what and when. Second lens — the numeric-and-structured-data check: Compare figures, units, groups, time windows, codes, ordered answers, signatures and references separately from prose. The correct target-language sentence can still be dangerous if an adjacent field contains a copied numerical error. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 076: The revised safety letter supersedes an earlier letter.

Controlled source excerpt (fictional): “The revised safety letter supersedes an earlier letter.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Both letters can be used interchangeably.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The specified new letter replaces the former for its stated purpose. Version confusion can result in obsolete participant information. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Match reference numbers and recipients. Second lens — the comprehension check: Ask the intended audience to explain what they understand without prompting. They should be able to describe the relevant choice or evidence in ordinary words. Misunderstanding is diagnostic feedback, not a reason to simplify away a right or a risk. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 077: A potential signal requires additional assessment.

Controlled source excerpt (fictional): “A potential signal requires additional assessment.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The signal proves an adverse drug reaction.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The signal warrants investigation without proving causal connection. Premature causality changes scientific and participant interpretation. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Distinguish signal from confirmed finding. Second lens — the cross-document check: Look for the same concept in the information sheet, protocol, schedule, form, registry and results text. If one version describes a different action or population, establish whether the source documents legitimately differ or whether the translation has drifted. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 078: Safety information is distributed to investigators.

Controlled source excerpt (fictional): “Safety information is distributed to investigators.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Every participant has already been told.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The distribution was to the named investigators as recorded. Recipients and completed communications cannot be inferred. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check distribution logs and participant notices. Second lens — the release-state check: A finished translation is not automatically approved for use. Verify scientific review, language review, required approvals, version status and distribution scope, following the study’s governance and applicable requirements. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 079: A monitoring visit found an unresolved issue.

Controlled source excerpt (fictional): “A monitoring visit found an unresolved issue.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The issue has been closed.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The issue remains unresolved as of the review. Closure is a different state requiring evidence. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Preserve issue ID and action owner. Second lens — the reader-decision check: Picture a person who never sees the original. Would this wording cause them to consent, decline, report, stop, sign, attend, share information or expect a benefit on a different basis? If so, the divergence is substantive even when grammar is flawless. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 080: The reported deaths occurred in the named analysis population.

Controlled source excerpt (fictional): “The reported deaths occurred in the named analysis population.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The treatment caused the listed deaths.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Deaths are reported in that population without adding attribution beyond the source. A count cannot be interpreted as causation from wording alone. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Confirm denominator and follow-up window. Second lens — the evidence-status check: Identify whether the phrase reports a plan, a possibility, an observation, a provisional assessment, an approved decision or a final result. Preserve that evidence level. Do not use later information to make the source sound more certain. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Protocol amendments, deviations and governance

Clinical research generates versions, approvals and deviations. The translator must preserve whether a sentence describes a plan, an approved change, an observation of noncompliance or an investigator’s permitted decision.

Diagnostic 081: Protocol amendment 4 changes the schedule from the stated effective date.

Controlled source excerpt (fictional): “Protocol amendment 4 changes the schedule from the stated effective date.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Every historical visit followed amendment 4.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The changed schedule applies according to its approved implementation terms. Amendments are not retroactive descriptions of older visits. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Compare version history. Second lens — the role-and-authority check: Identify the actor named in the controlled record. A study nurse, participant, investigator, sponsor, ethics committee and data system do different things. A translation that transfers power to another actor can be wrong without altering any technical noun. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 082: A deviation was documented during the stated visit.

Controlled source excerpt (fictional): “A deviation was documented during the stated visit.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The protocol was formally amended at that visit.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: A documented departure from the protocol occurred. Deviation and authorised amendment have different meanings. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Review original deviation log. Second lens — the scope-and-exception check: Underline every qualifier: if, unless, only, at least, no later than, optional, possible, expected, in the named group or under the stated circumstances. Test whether the target sentence covers exactly the same cases. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 083: The ethics committee approved the specified consent document version.

Controlled source excerpt (fictional): “The ethics committee approved the specified consent document version.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “All later translations are automatically approved.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Approval covers the identified document and version according to the record. An approval does not silently extend to changed text. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Verify approval identifiers. Second lens — the source-version check: Record the document identifier and revision. Match the line to the currently authorised study document; do not repair the scientific source by intuition. If the source conflicts with another controlled file, open a clarification ticket. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 084: The sponsor submitted an amendment for review.

Controlled source excerpt (fictional): “The sponsor submitted an amendment for review.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The amendment is already authorised everywhere.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The amendment was submitted, with approval status separately tracked. Submission is not final approval. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Inspect jurisdiction-specific status only for the actual study. Second lens — the comprehension check: Ask the intended audience to explain what they understand without prompting. They should be able to describe the relevant choice or evidence in ordinary words. Misunderstanding is diagnostic feedback, not a reason to simplify away a right or a risk. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 085: A corrective action has been proposed.

Controlled source excerpt (fictional): “A corrective action has been proposed.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The corrective action has been completed.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The action is proposed and not yet recorded as implemented. Proposals and completion states must remain distinct. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Compare CAPA statuses. Second lens — the escalation check: Some ambiguity is not translatable until the source owner answers it. Record the question, candidate meanings, clinical impact, responsible reviewer and authorised resolution. A silent guess is not quality assurance. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 086: The protocol defines a window for missed visits.

Controlled source excerpt (fictional): “The protocol defines a window for missed visits.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Any missed visit can be ignored.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The protocol specifies how a missed visit is handled. Do not replace a controlled process with a casual waiver. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check deviation and rescheduling instructions. Second lens — the cross-document check: Look for the same concept in the information sheet, protocol, schedule, form, registry and results text. If one version describes a different action or population, establish whether the source documents legitimately differ or whether the translation has drifted. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 087: A site-specific document adapts contact details only.

Controlled source excerpt (fictional): “A site-specific document adapts contact details only.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The site may rewrite participant rights freely.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Only the specified local fields are adapted within the authorised process. Localisation has limits set by governance. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Compare master and approved local versions. Second lens — the release-state check: A finished translation is not automatically approved for use. Verify scientific review, language review, required approvals, version status and distribution scope, following the study’s governance and applicable requirements. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 088: The investigator delegated a task to qualified staff.

Controlled source excerpt (fictional): “The investigator delegated a task to qualified staff.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The task no longer requires oversight.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: A delegated task remains within the recorded supervision framework. Delegation does not inherently remove accountability. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check delegation log and applicable procedures. Second lens — the reader-decision check: Picture a person who never sees the original. Would this wording cause them to consent, decline, report, stop, sign, attend, share information or expect a benefit on a different basis? If so, the divergence is substantive even when grammar is flawless. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 089: Monitoring findings are pending verification.

Controlled source excerpt (fictional): “Monitoring findings are pending verification.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The site has been found guilty of misconduct.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The findings are not yet confirmed as stated. Preliminary findings must not become final accusations. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Verify status of the review. Second lens — the evidence-status check: Identify whether the phrase reports a plan, a possibility, an observation, a provisional assessment, an approved decision or a final result. Preserve that evidence level. Do not use later information to make the source sound more certain. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 090: An approved deviation does not change the endpoint definition.

Controlled source excerpt (fictional): “An approved deviation does not change the endpoint definition.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The endpoint has been changed by exception.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The deviation and predefined outcome remain separate records. A one-off event should not silently redefine what was measured. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Trace protocol and analysis plan. Second lens — the role-and-authority check: Identify the actor named in the controlled record. A study nurse, participant, investigator, sponsor, ethics committee and data system do different things. A translation that transfers power to another actor can be wrong without altering any technical noun. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Outcome measures and statistical reporting

Outcomes are not simply good or bad news. Their interpretation depends on definition, group, time, statistical model, missing data and uncertainty. Translating a result into a headline is a separate editorial act that must not rewrite the evidence.

Diagnostic 091: The primary endpoint was measured at the specified week.

Controlled source excerpt (fictional): “The primary endpoint was measured at the specified week.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The study’s primary purpose was achieved at that week.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The primary outcome was measured at the stated time point. Measurement does not establish success. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check endpoint definition and numerical result. Second lens — the source-version check: Record the document identifier and revision. Match the line to the currently authorised study document; do not repair the scientific source by intuition. If the source conflicts with another controlled file, open a clarification ticket. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 092: The between-group difference was estimated with an interval.

Controlled source excerpt (fictional): “The between-group difference was estimated with an interval.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The difference is exactly equal to the point estimate for everyone.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Report the estimate with its uncertainty interval. Point estimates are not individual guarantees. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Keep confidence interval bounds and direction. Second lens — the numeric-and-structured-data check: Compare figures, units, groups, time windows, codes, ordered answers, signatures and references separately from prose. The correct target-language sentence can still be dangerous if an adjacent field contains a copied numerical error. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 093: The analysis did not demonstrate superiority.

Controlled source excerpt (fictional): “The analysis did not demonstrate superiority.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The two treatments are proven identical.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The specified analysis did not establish superiority. Failure to show superiority does not prove equivalence. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check test objective and limits. Second lens — the comprehension check: Ask the intended audience to explain what they understand without prompting. They should be able to describe the relevant choice or evidence in ordinary words. Misunderstanding is diagnostic feedback, not a reason to simplify away a right or a risk. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 094: An exploratory subgroup appeared to differ.

Controlled source excerpt (fictional): “An exploratory subgroup appeared to differ.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The treatment definitely works for that subgroup.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: An exploratory subgroup analysis suggested a possible difference. Exploratory findings may require further confirmation. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Preserve subgroup and exploratory qualifiers. Second lens — the escalation check: Some ambiguity is not translatable until the source owner answers it. Record the question, candidate meanings, clinical impact, responsible reviewer and authorised resolution. A silent guess is not quality assurance. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 095: The risk ratio was below one for the defined comparison.

Controlled source excerpt (fictional): “The risk ratio was below one for the defined comparison.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The risk is zero.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The estimated relative measure is below one, with the stated context. A ratio below one is not absence of events. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Keep numerator, denominator and uncertainty. Second lens — the cross-document check: Look for the same concept in the information sheet, protocol, schedule, form, registry and results text. If one version describes a different action or population, establish whether the source documents legitimately differ or whether the translation has drifted. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 096: The median follow-up was stated in months.

Controlled source excerpt (fictional): “The median follow-up was stated in months.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Every participant was followed for exactly that duration.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The median follow-up was the reported value. A population summary is not an individual schedule. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check metric and unit. Second lens — the release-state check: A finished translation is not automatically approved for use. Verify scientific review, language review, required approvals, version status and distribution scope, following the study’s governance and applicable requirements. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 097: Missing outcome data were handled with a specified method.

Controlled source excerpt (fictional): “Missing outcome data were handled with a specified method.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Missing data did not exist.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The report describes the named approach to missing observations. An analysis method is not proof of complete data. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Retain method name and limitations. Second lens — the reader-decision check: Picture a person who never sees the original. Would this wording cause them to consent, decline, report, stop, sign, attend, share information or expect a benefit on a different basis? If so, the divergence is substantive even when grammar is flawless. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 098: Statistical significance was observed for a specified comparison.

Controlled source excerpt (fictional): “Statistical significance was observed for a specified comparison.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The improvement is necessarily important to every patient.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The specified comparison met its statistical criterion. Statistical and clinical importance are different questions. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check effect magnitude and relevance. Second lens — the role-and-authority check: Identify the actor named in the controlled record. A study nurse, participant, investigator, sponsor, ethics committee and data system do different things. A translation that transfers power to another actor can be wrong without altering any technical noun. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 099: The interval includes the null value.

Controlled source excerpt (fictional): “The interval includes the null value.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The evidence establishes a definite effect.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The reported interval includes the reference value for no difference as defined. Interpretation must reflect the uncertainty shown. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Verify direction and interval scale. Second lens — the scope-and-exception check: Underline every qualifier: if, unless, only, at least, no later than, optional, possible, expected, in the named group or under the stated circumstances. Test whether the target sentence covers exactly the same cases. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 100: The study met its prespecified endpoint under the analysis plan.

Controlled source excerpt (fictional): “The study met its prespecified endpoint under the analysis plan.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “All hypotheses and all groups were validated.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The named endpoint was met under the stated analysis. A single success does not establish every related claim. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Compare prespecified and secondary outcomes. Second lens — the source-version check: Record the document identifier and revision. Match the line to the currently authorised study document; do not repair the scientific source by intuition. If the source conflicts with another controlled file, open a clarification ticket. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Participant reimbursement and compensation

Money words can alter consent. Reimbursement for a cost, payment for time, compensation for injury and financial inducement are not synonyms. Translators must preserve condition, amount, schedule, responsibility and local approval.

Diagnostic 101: Reasonable travel costs may be reimbursed with receipts.

Controlled source excerpt (fictional): “Reasonable travel costs may be reimbursed with receipts.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Travel is always free in advance.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Eligible travel expenses may be reimbursed under the stated process. Reimbursement is not automatic prepaid transport. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check criteria and claim method. Second lens — the numeric-and-structured-data check: Compare figures, units, groups, time windows, codes, ordered answers, signatures and references separately from prose. The correct target-language sentence can still be dangerous if an adjacent field contains a copied numerical error. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 102: Participants receive payment for completed visits as described.

Controlled source excerpt (fictional): “Participants receive payment for completed visits as described.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Participants receive the full amount upon signing.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The approved payment is linked to the stated participation milestones. A payment schedule can affect expectations. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Verify the table and conditions. Second lens — the comprehension check: Ask the intended audience to explain what they understand without prompting. They should be able to describe the relevant choice or evidence in ordinary words. Misunderstanding is diagnostic feedback, not a reason to simplify away a right or a risk. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 103: You may withdraw without losing payments already earned.

Controlled source excerpt (fictional): “You may withdraw without losing payments already earned.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Withdrawing means all past payments are forfeited.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Withdrawal does not forfeit the stated accrued payments. The incorrect version creates pressure to remain enrolled. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check payment and withdrawal clauses together. Second lens — the escalation check: Some ambiguity is not translatable until the source owner answers it. Record the question, candidate meanings, clinical impact, responsible reviewer and authorised resolution. A silent guess is not quality assurance. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 104: The study will not charge for specified research procedures.

Controlled source excerpt (fictional): “The study will not charge for specified research procedures.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “All future healthcare is permanently free.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Only the identified study-related procedures are covered as stated. A bounded financial statement must not become lifetime insurance. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check covered services. Second lens — the cross-document check: Look for the same concept in the information sheet, protocol, schedule, form, registry and results text. If one version describes a different action or population, establish whether the source documents legitimately differ or whether the translation has drifted. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 105: The insurer may provide compensation under defined arrangements.

Controlled source excerpt (fictional): “The insurer may provide compensation under defined arrangements.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Every injury is automatically compensated in full.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Compensation follows the described arrangements and conditions. Do not invent a guarantee or deny a possible pathway. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Verify claims contact and policy language. Second lens — the reader-decision check: Picture a person who never sees the original. Would this wording cause them to consent, decline, report, stop, sign, attend, share information or expect a benefit on a different basis? If so, the divergence is substantive even when grammar is flawless. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 106: An inconvenience allowance may be available.

Controlled source excerpt (fictional): “An inconvenience allowance may be available.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The payment is a guaranteed medical benefit.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The allowance is compensation for time or inconvenience as described. Payment is not a treatment outcome. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Keep financial and clinical sections separate. Second lens — the evidence-status check: Identify whether the phrase reports a plan, a possibility, an observation, a provisional assessment, an approved decision or a final result. Preserve that evidence level. Do not use later information to make the source sound more certain. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 107: The sponsor pays the research site’s costs.

Controlled source excerpt (fictional): “The sponsor pays the research site’s costs.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The participant receives that entire amount.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The cost is paid to the named institution or service. Recipient matters as much as amount. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check payer and payee fields. Second lens — the role-and-authority check: Identify the actor named in the controlled record. A study nurse, participant, investigator, sponsor, ethics committee and data system do different things. A translation that transfers power to another actor can be wrong without altering any technical noun. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 108: The participant may be asked to provide receipts.

Controlled source excerpt (fictional): “The participant may be asked to provide receipts.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “No documentation is ever required.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Supporting receipts may be required under the stated rules. Dropping a condition can frustrate a later claim. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Keep documentation requirements accessible. Second lens — the scope-and-exception check: Underline every qualifier: if, unless, only, at least, no later than, optional, possible, expected, in the named group or under the stated circumstances. Test whether the target sentence covers exactly the same cases. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 109: The payment is stated in a specified currency.

Controlled source excerpt (fictional): “The payment is stated in a specified currency.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The numerical amount is unchanged when the currency label changes.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Preserve currency and amount, or document any authorised conversion clearly. Uncontrolled currency changes are factual errors. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Double-check symbol, currency code and conversion date. Second lens — the source-version check: Record the document identifier and revision. Match the line to the currently authorised study document; do not repair the scientific source by intuition. If the source conflicts with another controlled file, open a clarification ticket. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 110: Payment will not depend on a favourable medical result.

Controlled source excerpt (fictional): “Payment will not depend on a favourable medical result.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “You are paid only if the product works.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Payment is independent of the stated clinical outcome. An outcome-contingent reward changes the moral and practical meaning. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Read the approved payment explanation aloud. Second lens — the numeric-and-structured-data check: Compare figures, units, groups, time windows, codes, ordered answers, signatures and references separately from prose. The correct target-language sentence can still be dangerous if an adjacent field contains a copied numerical error. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Privacy, data access and future specimen use

A participant’s data and samples may have different governance. Translators should not promise absolute anonymity where records remain linkable or assume that future research is automatically included.

Diagnostic 111: Your records will use a study code.

Controlled source excerpt (fictional): “Your records will use a study code.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “No person anywhere can reconnect the record to you.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The record is coded, with access to the link controlled as described. Coding is not identical to irreversible anonymisation. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Confirm the re-identification arrangements. Second lens — the comprehension check: Ask the intended audience to explain what they understand without prompting. They should be able to describe the relevant choice or evidence in ordinary words. Misunderstanding is diagnostic feedback, not a reason to simplify away a right or a risk. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 112: Only authorised study personnel may access specified records.

Controlled source excerpt (fictional): “Only authorised study personnel may access specified records.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Anyone involved in health care can see everything.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Access is limited according to the approved description. Access scope is a trust-critical fact. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Compare the role and data lists. Second lens — the cross-document check: Look for the same concept in the information sheet, protocol, schedule, form, registry and results text. If one version describes a different action or population, establish whether the source documents legitimately differ or whether the translation has drifted. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 113: Data may be transferred to named recipients for study purposes.

Controlled source excerpt (fictional): “Data may be transferred to named recipients for study purposes.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Data will never leave the clinic.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The specified information may be shared with the stated recipients. False secrecy and excessive openness are both inaccurate. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check the data flow notice. Second lens — the release-state check: A finished translation is not automatically approved for use. Verify scientific review, language review, required approvals, version status and distribution scope, following the study’s governance and applicable requirements. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 114: Your identifiable information will be retained for the stated period.

Controlled source excerpt (fictional): “Your identifiable information will be retained for the stated period.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “All your data disappear as soon as the study ends.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Retention follows the described periods and categories. Consent to storage should not be based on false deletion promises. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check retention rules by data type. Second lens — the reader-decision check: Picture a person who never sees the original. Would this wording cause them to consent, decline, report, stop, sign, attend, share information or expect a benefit on a different basis? If so, the divergence is substantive even when grammar is flawless. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 115: Your optional samples may be used in a future study if you agree.

Controlled source excerpt (fictional): “Your optional samples may be used in a future study if you agree.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Future sample research is compulsory.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Optional future use requires the separate choice described. An optional permission must not hide inside general consent. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Test checkbox logic. Second lens — the evidence-status check: Identify whether the phrase reports a plan, a possibility, an observation, a provisional assessment, an approved decision or a final result. Preserve that evidence level. Do not use later information to make the source sound more certain. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 116: De-identified results may be published in aggregate.

Controlled source excerpt (fictional): “De-identified results may be published in aggregate.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Your full name will be published with the summary.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Aggregate results do not automatically reveal individual identities as stated. The audience’s privacy inference changes with the wording. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check actual publication plan. Second lens — the role-and-authority check: Identify the actor named in the controlled record. A study nurse, participant, investigator, sponsor, ethics committee and data system do different things. A translation that transfers power to another actor can be wrong without altering any technical noun. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 117: A participant may request information about data rights.

Controlled source excerpt (fictional): “A participant may request information about data rights.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “All requests for deletion must be granted immediately.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Participants may inquire about the rights and limits described in the study information. Legal rights vary; do not invent a universal unconditional entitlement. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Preserve jurisdiction-specific legal references only when applicable. Second lens — the scope-and-exception check: Underline every qualifier: if, unless, only, at least, no later than, optional, possible, expected, in the named group or under the stated circumstances. Test whether the target sentence covers exactly the same cases. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 118: A biological specimen is stored under a coded identifier.

Controlled source excerpt (fictional): “A biological specimen is stored under a coded identifier.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The specimen is biologically anonymous in every context.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The specimen is stored with the stated coding and access controls. A code does not eliminate every possible identification concern. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check specimen-label policy. Second lens — the source-version check: Record the document identifier and revision. Match the line to the currently authorised study document; do not repair the scientific source by intuition. If the source conflicts with another controlled file, open a clarification ticket. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 119: Research data may undergo quality review.

Controlled source excerpt (fictional): “Research data may undergo quality review.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Every employee may use the data for unrelated purposes.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Quality review occurs within the described permissions and safeguards. An authorised quality activity is not a blanket reuse licence. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check scope and purpose limitation. Second lens — the comprehension check: Ask the intended audience to explain what they understand without prompting. They should be able to describe the relevant choice or evidence in ordinary words. Misunderstanding is diagnostic feedback, not a reason to simplify away a right or a risk. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 120: Withdrawal from future sample use is subject to the approved limitations.

Controlled source excerpt (fictional): “Withdrawal from future sample use is subject to the approved limitations.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “All previous sample analyses can always be reversed.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The source explains what uses can still be stopped and what may be irreversible. Overpromising reversal creates false expectations. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Review the precise withdrawal paragraph. Second lens — the escalation check: Some ambiguity is not translatable until the source owner answers it. Record the question, candidate meanings, clinical impact, responsible reviewer and authorised resolution. A silent guess is not quality assurance. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Questionnaires, patient-reported outcomes and scales

A survey can be grammatically accurate yet statistically changed by its translation. Preserve response anchors, recall periods, instructions, skip logic, register and construct equivalence; do not treat a validated instrument as free-form copy.

Diagnostic 121: During the past seven days, how often did you feel tired?

Controlled source excerpt (fictional): “During the past seven days, how often did you feel tired?” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “How tired do you feel right now?” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Ask about frequency during the last seven days. Changing recall period and variable destroys comparability. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check response anchors and item specification. Second lens — the cross-document check: Look for the same concept in the information sheet, protocol, schedule, form, registry and results text. If one version describes a different action or population, establish whether the source documents legitimately differ or whether the translation has drifted. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 122: Never, rarely, sometimes, often, always.

Controlled source excerpt (fictional): “Never, rarely, sometimes, often, always.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Always, often, sometimes, rarely, never without updating codes.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Keep the intended anchor order and coded values. A reversed scale changes data interpretation. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Test each mapped code. Second lens — the release-state check: A finished translation is not automatically approved for use. Verify scientific review, language review, required approvals, version status and distribution scope, following the study’s governance and applicable requirements. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 123: Rate your pain at its worst today.

Controlled source excerpt (fictional): “Rate your pain at its worst today.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Rate your average pain this month.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Ask about worst pain during the specified day. Statistic and recall window both matter. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Review item definition. Second lens — the reader-decision check: Picture a person who never sees the original. Would this wording cause them to consent, decline, report, stop, sign, attend, share information or expect a benefit on a different basis? If so, the divergence is substantive even when grammar is flawless. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 124: Select all answers that apply.

Controlled source excerpt (fictional): “Select all answers that apply.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Choose exactly one answer.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Allow multiple selections as stated. Changing input control changes data, not just language. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Test the form interface. Second lens — the evidence-status check: Identify whether the phrase reports a plan, a possibility, an observation, a provisional assessment, an approved decision or a final result. Preserve that evidence level. Do not use later information to make the source sound more certain. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 125: If you answered no, skip the next question.

Controlled source excerpt (fictional): “If you answered no, skip the next question.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Everyone must answer the next question.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Retain the stated conditional branching logic. Skip logic is executable meaning. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Simulate both response paths. Second lens — the role-and-authority check: Identify the actor named in the controlled record. A study nurse, participant, investigator, sponsor, ethics committee and data system do different things. A translation that transfers power to another actor can be wrong without altering any technical noun. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 126: A response of not applicable is available.

Controlled source excerpt (fictional): “A response of not applicable is available.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Missing answers must be recorded as no.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Keep not applicable separate from no and missing. These categories carry different evidence. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check survey export mapping. Second lens — the source-version check: Record the document identifier and revision. Match the line to the currently authorised study document; do not repair the scientific source by intuition. If the source conflicts with another controlled file, open a clarification ticket. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 127: The item asks about interference with work or daily activities.

Controlled source excerpt (fictional): “The item asks about interference with work or daily activities.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The item asks whether the participant is employed.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Ask about impact on the defined activities, not employment status. A proxy variable is not the original construct. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Discuss cultural interpretation with instrument owners. Second lens — the numeric-and-structured-data check: Compare figures, units, groups, time windows, codes, ordered answers, signatures and references separately from prose. The correct target-language sentence can still be dangerous if an adjacent field contains a copied numerical error. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 128: The questionnaire requires use of an approved translated version.

Controlled source excerpt (fictional): “The questionnaire requires use of an approved translated version.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Any colloquial paraphrase is acceptable.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Use the validated or authorised language version when the study requires it. Uncontrolled wording can break measurement validity. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check licensing and version instructions. Second lens — the comprehension check: Ask the intended audience to explain what they understand without prompting. They should be able to describe the relevant choice or evidence in ordinary words. Misunderstanding is diagnostic feedback, not a reason to simplify away a right or a risk. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 129: The scale uses a numerical score with labelled endpoints.

Controlled source excerpt (fictional): “The scale uses a numerical score with labelled endpoints.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Each number can be reordered to look nicer.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Retain the ordered scale and its endpoint descriptions. Visual layout is part of measurement meaning. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Test mobile and paper formats. Second lens — the escalation check: Some ambiguity is not translatable until the source owner answers it. Record the question, candidate meanings, clinical impact, responsible reviewer and authorised resolution. A silent guess is not quality assurance. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 130: A patient-reported outcome is reported by the participant.

Controlled source excerpt (fictional): “A patient-reported outcome is reported by the participant.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The clinician fills in every personal answer.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The specified response comes directly from the participant. Source of information matters for analysis. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Verify reporter identity in the form. Second lens — the cross-document check: Look for the same concept in the information sheet, protocol, schedule, form, registry and results text. If one version describes a different action or population, establish whether the source documents legitimately differ or whether the translation has drifted. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Public registries, scientific summaries and results

The trial registry record, technical results report, media summary and participant-language summary have different audiences but need a shared evidentiary spine. Translating across those genres should not inflate the results.

Diagnostic 131: The registry lists the study as active but not recruiting.

Controlled source excerpt (fictional): “The registry lists the study as active but not recruiting.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The trial is actively recruiting everyone.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The study remains active while new recruitment is closed. Status labels are not marketing synonyms. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check the dated public record. Second lens — the release-state check: A finished translation is not automatically approved for use. Verify scientific review, language review, required approvals, version status and distribution scope, following the study’s governance and applicable requirements. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 132: A primary outcome measure is defined before results are available.

Controlled source excerpt (fictional): “A primary outcome measure is defined before results are available.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The primary outcome proves the intervention works.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The named primary measure is a pre-specified way to assess an outcome. An endpoint definition is not evidence of success. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Compare protocol and registration. Second lens — the reader-decision check: Picture a person who never sees the original. Would this wording cause them to consent, decline, report, stop, sign, attend, share information or expect a benefit on a different basis? If so, the divergence is substantive even when grammar is flawless. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 133: The reported results are preliminary.

Controlled source excerpt (fictional): “The reported results are preliminary.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The findings are final and conclusive.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The results are preliminary as described. Tentative evidence should not be marketed as settled science. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check data lock and analysis status. Second lens — the role-and-authority check: Identify the actor named in the controlled record. A study nurse, participant, investigator, sponsor, ethics committee and data system do different things. A translation that transfers power to another actor can be wrong without altering any technical noun. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 134: Results were posted after completion.

Controlled source excerpt (fictional): “Results were posted after completion.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “All people who signed up completed the study.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: A publication milestone does not describe participant retention. Do not infer flow counts from posting status. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Read the participant-flow table. Second lens — the scope-and-exception check: Underline every qualifier: if, unless, only, at least, no later than, optional, possible, expected, in the named group or under the stated circumstances. Test whether the target sentence covers exactly the same cases. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 135: Adverse event results are presented by treatment group.

Controlled source excerpt (fictional): “Adverse event results are presented by treatment group.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Every event was caused by each treatment.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Events are tabulated by the specified assigned group without an automatic causality claim. Group attribution is not event causation. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check table headers. Second lens — the source-version check: Record the document identifier and revision. Match the line to the currently authorised study document; do not repair the scientific source by intuition. If the source conflicts with another controlled file, open a clarification ticket. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 136: The sample size refers to participants analysed.

Controlled source excerpt (fictional): “The sample size refers to participants analysed.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The number is everyone who expressed interest.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The sample size refers to the stated analysis population. Recruitment interest, enrolment and analysis are different counts. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Preserve denominator label. Second lens — the numeric-and-structured-data check: Compare figures, units, groups, time windows, codes, ordered answers, signatures and references separately from prose. The correct target-language sentence can still be dangerous if an adjacent field contains a copied numerical error. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 137: A peer-reviewed paper discusses the trial.

Controlled source excerpt (fictional): “A peer-reviewed paper discusses the trial.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Peer review means the intervention has regulatory approval.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The study was reported in a peer-reviewed publication. Publication review is not product authorisation. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check whether regulatory status is stated separately. Second lens — the comprehension check: Ask the intended audience to explain what they understand without prompting. They should be able to describe the relevant choice or evidence in ordinary words. Misunderstanding is diagnostic feedback, not a reason to simplify away a right or a risk. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 138: The registry record was updated on a given date.

Controlled source excerpt (fictional): “The registry record was updated on a given date.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The intervention was administered on that date.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The date is a registry update timestamp. Publication timing should not become procedure timing. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Identify field provenance. Second lens — the escalation check: Some ambiguity is not translatable until the source owner answers it. Record the question, candidate meanings, clinical impact, responsible reviewer and authorised resolution. A silent guess is not quality assurance. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 139: The lay summary uses plain language to report limitations.

Controlled source excerpt (fictional): “The lay summary uses plain language to report limitations.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The lay summary can delete all negative or uncertain findings.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The summary should preserve the study’s material findings and limitations. Plain language does not justify omission of balancing evidence. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Compare technical and lay summaries. Second lens — the cross-document check: Look for the same concept in the information sheet, protocol, schedule, form, registry and results text. If one version describes a different action or population, establish whether the source documents legitimately differ or whether the translation has drifted. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 140: An outcome is reported as not measured in this analysis.

Controlled source excerpt (fictional): “An outcome is reported as not measured in this analysis.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The outcome was zero.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The source indicates that this analysis does not report the measure. Missing or unmeasured is not numerically zero. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check data tables and report notes. Second lens — the reader-decision check: Picture a person who never sees the original. Would this wording cause them to consent, decline, report, stop, sign, attend, share information or expect a benefit on a different basis? If so, the divergence is substantive even when grammar is flawless. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Digital eConsent, remote visits and multilingual systems

Electronic consent is a coupled language-and-interface problem. The words, branch conditions, display order, signature records and accessibility affordances together determine what the reader actually sees.

Diagnostic 141: Click continue to view the next information section.

Controlled source excerpt (fictional): “Click continue to view the next information section.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Click agree to enrol immediately.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Continue opens the next section without recording consent. Navigation and consent actions are different. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Test the button state and stored event. Second lens — the evidence-status check: Identify whether the phrase reports a plan, a possibility, an observation, a provisional assessment, an approved decision or a final result. Preserve that evidence level. Do not use later information to make the source sound more certain. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 142: The participant can download a copy before signing.

Controlled source excerpt (fictional): “The participant can download a copy before signing.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “A copy is available only after study completion.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The record is available at the specified stage for participant review. Access timing affects deliberation. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Test the download workflow. Second lens — the role-and-authority check: Identify the actor named in the controlled record. A study nurse, participant, investigator, sponsor, ethics committee and data system do different things. A translation that transfers power to another actor can be wrong without altering any technical noun. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 143: The electronic signature records the agreed document version.

Controlled source excerpt (fictional): “The electronic signature records the agreed document version.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The signature applies to every future amendment.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The signature applies to the displayed controlled version. A signature cannot automatically cover unseen future wording. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check version hash or document ID. Second lens — the scope-and-exception check: Underline every qualifier: if, unless, only, at least, no later than, optional, possible, expected, in the named group or under the stated circumstances. Test whether the target sentence covers exactly the same cases. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 144: A pop-up includes an important risk qualification.

Controlled source excerpt (fictional): “A pop-up includes an important risk qualification.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The qualification is omitted on mobile.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The risk information must remain perceivable in the released interface. Hidden display content is a translation failure. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Test narrow screens and assistive technology. Second lens — the source-version check: Record the document identifier and revision. Match the line to the currently authorised study document; do not repair the scientific source by intuition. If the source conflicts with another controlled file, open a clarification ticket. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 145: The language selector changes explanatory text.

Controlled source excerpt (fictional): “The language selector changes explanatory text.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The selector changes the underlying study randomisation.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Only authorised display text changes; study data controls remain stable. Language preference cannot corrupt treatment assignment. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Inspect localisation keys and backend behaviour. Second lens — the numeric-and-structured-data check: Compare figures, units, groups, time windows, codes, ordered answers, signatures and references separately from prose. The correct target-language sentence can still be dangerous if an adjacent field contains a copied numerical error. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 146: A conditional question appears when a participant chooses yes.

Controlled source excerpt (fictional): “A conditional question appears when a participant chooses yes.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The question appears for everyone regardless of answer.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The branch follows the prescribed response condition. Translation must respect executable skip logic. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Test both user pathways. Second lens — the comprehension check: Ask the intended audience to explain what they understand without prompting. They should be able to describe the relevant choice or evidence in ordinary words. Misunderstanding is diagnostic feedback, not a reason to simplify away a right or a risk. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 147: Remote consent may be used under approved conditions.

Controlled source excerpt (fictional): “Remote consent may be used under approved conditions.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Remote consent is always permitted everywhere.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Remote consent follows the stated authorised circumstances. Technology availability is not a legal approval. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check study and local governance. Second lens — the cross-document check: Look for the same concept in the information sheet, protocol, schedule, form, registry and results text. If one version describes a different action or population, establish whether the source documents legitimately differ or whether the translation has drifted. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 148: An optional consent item is unchecked by default.

Controlled source excerpt (fictional): “An optional consent item is unchecked by default.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The item is preselected as yes.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The participant must be able to choose the optional answer as designed. Preselection can bias consent choices. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Verify form defaults and accessibility. Second lens — the release-state check: A finished translation is not automatically approved for use. Verify scientific review, language review, required approvals, version status and distribution scope, following the study’s governance and applicable requirements. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 149: A recorded call supplements the consent process where permitted.

Controlled source excerpt (fictional): “A recorded call supplements the consent process where permitted.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “All consent discussions must be recorded.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Only the approved recording practice applies. The option or requirement cannot be invented. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check participant notice and privacy policy. Second lens — the reader-decision check: Picture a person who never sees the original. Would this wording cause them to consent, decline, report, stop, sign, attend, share information or expect a benefit on a different basis? If so, the divergence is substantive even when grammar is flawless. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 150: A failed upload produces an error without saving the signature.

Controlled source excerpt (fictional): “A failed upload produces an error without saving the signature.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The screen says consent is complete despite the failure.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The interface should accurately reflect the recorded state of the signature. A success message can become false evidence. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Test error states in each language. Second lens — the evidence-status check: Identify whether the phrase reports a plan, a possibility, an observation, a provisional assessment, an approved decision or a final result. Preserve that evidence level. Do not use later information to make the source sound more certain. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Interpreting, accessibility and comprehension

Human access to the study information cannot be inferred from a language menu alone. Spoken interpretation, literacy support, sensory accessibility, terminology and participant comprehension may require separate workstreams.

Diagnostic 151: An interpreter can assist with the consent discussion as arranged.

Controlled source excerpt (fictional): “An interpreter can assist with the consent discussion as arranged.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The printed translation replaces all discussion.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: A qualified interpreter supports the discussion when needed and authorised. Written documents and live conversation have different functions. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check the approved communication plan. Second lens — the role-and-authority check: Identify the actor named in the controlled record. A study nurse, participant, investigator, sponsor, ethics committee and data system do different things. A translation that transfers power to another actor can be wrong without altering any technical noun. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 152: The participant may request clarification in their preferred language.

Controlled source excerpt (fictional): “The participant may request clarification in their preferred language.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Every language is guaranteed immediately at all sites.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The source offers the described clarification route, without invented availability promises. A service claim can fail in practice. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Verify real service capability. Second lens — the scope-and-exception check: Underline every qualifier: if, unless, only, at least, no later than, optional, possible, expected, in the named group or under the stated circumstances. Test whether the target sentence covers exactly the same cases. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 153: A witness records the specified consent process when required.

Controlled source excerpt (fictional): “A witness records the specified consent process when required.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The witness decides which treatment is clinically best.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The witness fulfils the recorded procedural role. Witnessing is not medical decision-making. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Preserve role boundaries. Second lens — the source-version check: Record the document identifier and revision. Match the line to the currently authorised study document; do not repair the scientific source by intuition. If the source conflicts with another controlled file, open a clarification ticket. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 154: A simplified explanation accompanies the formal term.

Controlled source excerpt (fictional): “A simplified explanation accompanies the formal term.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The formal risk term may be deleted entirely.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Keep the controlling concept while explaining it accessibly. Plain language should add clarity without removing meaning. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Compare accessible and technical versions. Second lens — the comprehension check: Ask the intended audience to explain what they understand without prompting. They should be able to describe the relevant choice or evidence in ordinary words. Misunderstanding is diagnostic feedback, not a reason to simplify away a right or a risk. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 155: A participant can use an accessible document format.

Controlled source excerpt (fictional): “A participant can use an accessible document format.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Only audio is permitted even when reading is preferred.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Provide the stated accessible format options. Accessibility choice should not be restricted by translation. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check screen reader and text alternatives. Second lens — the escalation check: Some ambiguity is not translatable until the source owner answers it. Record the question, candidate meanings, clinical impact, responsible reviewer and authorised resolution. A silent guess is not quality assurance. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 156: The consent discussion includes an opportunity to ask questions.

Controlled source excerpt (fictional): “The consent discussion includes an opportunity to ask questions.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “A comprehension quiz is the only form of consent.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The discussion permits questions and understanding checks as applicable. One test cannot replace the whole process. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Read the consent workflow. Second lens — the cross-document check: Look for the same concept in the information sheet, protocol, schedule, form, registry and results text. If one version describes a different action or population, establish whether the source documents legitimately differ or whether the translation has drifted. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 157: The translated document retains each numbered subsection.

Controlled source excerpt (fictional): “The translated document retains each numbered subsection.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Sections may be rearranged so rights are hidden at the end.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The document preserves essential information and an approved comprehensible order. Structure affects what a person notices first. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Compare information hierarchy. Second lens — the release-state check: A finished translation is not automatically approved for use. Verify scientific review, language review, required approvals, version status and distribution scope, following the study’s governance and applicable requirements. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 158: A language is chosen based on the participant’s understanding.

Controlled source excerpt (fictional): “A language is chosen based on the participant’s understanding.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Nationality automatically determines language competence.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Communication should use a language the participant actually understands. Assuming language from background can exclude people. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Confirm actual communication needs respectfully. Second lens — the reader-decision check: Picture a person who never sees the original. Would this wording cause them to consent, decline, report, stop, sign, attend, share information or expect a benefit on a different basis? If so, the divergence is substantive even when grammar is flawless. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 159: A technical term is explained with an example.

Controlled source excerpt (fictional): “A technical term is explained with an example.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The example can make an unapproved medical recommendation.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The explanation clarifies the term without adding treatment advice. Examples must not become new instructions. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Review every added example with the source owner. Second lens — the evidence-status check: Identify whether the phrase reports a plan, a possibility, an observation, a provisional assessment, an approved decision or a final result. Preserve that evidence level. Do not use later information to make the source sound more certain. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 160: A comprehension concern is referred to the responsible study personnel.

Controlled source excerpt (fictional): “A comprehension concern is referred to the responsible study personnel.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The translator signs for informed consent on the participant’s behalf.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The translation team flags concerns; authorised personnel conduct consent. Linguistic help does not confer consent authority. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check recorded responsibilities. Second lens — the role-and-authority check: Identify the actor named in the controlled record. A study nurse, participant, investigator, sponsor, ethics committee and data system do different things. A translation that transfers power to another actor can be wrong without altering any technical noun. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Clinical data forms, coding and machine-readable records

Research translation enters databases through case-report forms, structured questionnaires and terminology coding. Natural-language labels can be localised while identifiers and data types remain governed by their system specification.

Diagnostic 161: A case report form uses a stable field identifier.

Controlled source excerpt (fictional): “A case report form uses a stable field identifier.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The translated label replaces the database field key.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Keep the field identifier stable while adapting allowed display text. A key change can break data collection and analysis. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Compare export schema before and after localisation. Second lens — the source-version check: Record the document identifier and revision. Match the line to the currently authorised study document; do not repair the scientific source by intuition. If the source conflicts with another controlled file, open a clarification ticket. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 162: An adverse event code maps to a controlled terminology version.

Controlled source excerpt (fictional): “An adverse event code maps to a controlled terminology version.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The code may be translated into an invented local code.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Preserve controlled code identity while explaining the preferred term where authorised. Label and code are not interchangeable. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Verify dictionary version and code list. Second lens — the numeric-and-structured-data check: Compare figures, units, groups, time windows, codes, ordered answers, signatures and references separately from prose. The correct target-language sentence can still be dangerous if an adjacent field contains a copied numerical error. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 163: A dropdown value means not assessed.

Controlled source excerpt (fictional): “A dropdown value means not assessed.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The translation maps it to normal.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Not assessed must remain distinct from a negative or normal result. Data meaning changes when unknown becomes normal. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Test each enumerated value. Second lens — the comprehension check: Ask the intended audience to explain what they understand without prompting. They should be able to describe the relevant choice or evidence in ordinary words. Misunderstanding is diagnostic feedback, not a reason to simplify away a right or a risk. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 164: A date field uses an explicit year-month-day schema.

Controlled source excerpt (fictional): “A date field uses an explicit year-month-day schema.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The field order can change without conversion.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Display dates appropriately while preserving the underlying validated date value. Date order is a frequent source of silent errors. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Test ambiguous calendar dates. Second lens — the escalation check: Some ambiguity is not translatable until the source owner answers it. Record the question, candidate meanings, clinical impact, responsible reviewer and authorised resolution. A silent guess is not quality assurance. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 165: A decimal value uses the source measurement unit.

Controlled source excerpt (fictional): “A decimal value uses the source measurement unit.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The unit can be switched without converting the number.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Retain unit and number as an inseparable pair. Uncontrolled conversion corrupts numeric evidence. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Cross-check against original values. Second lens — the cross-document check: Look for the same concept in the information sheet, protocol, schedule, form, registry and results text. If one version describes a different action or population, establish whether the source documents legitimately differ or whether the translation has drifted. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 166: A missing value is distinct from zero.

Controlled source excerpt (fictional): “A missing value is distinct from zero.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “All blanks are automatically treated as zero.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Keep missing data separate from numeric zero. The analysis may treat the two states very differently. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Inspect data-entry rules. Second lens — the release-state check: A finished translation is not automatically approved for use. Verify scientific review, language review, required approvals, version status and distribution scope, following the study’s governance and applicable requirements. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 167: The form names an investigator-assessed outcome.

Controlled source excerpt (fictional): “The form names an investigator-assessed outcome.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The translated label says patient-reported outcome.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Keep the recorded assessor type. Who made the assessment is part of the evidence. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check field provenance. Second lens — the reader-decision check: Picture a person who never sees the original. Would this wording cause them to consent, decline, report, stop, sign, attend, share information or expect a benefit on a different basis? If so, the divergence is substantive even when grammar is flawless. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 168: An audit trail preserves prior data corrections.

Controlled source excerpt (fictional): “An audit trail preserves prior data corrections.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The translation replaces history with a single cleaned value.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Keep audit-history concepts and events separate from final displayed values. Traceability matters when data are revised. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Review audit export. Second lens — the role-and-authority check: Identify the actor named in the controlled record. A study nurse, participant, investigator, sponsor, ethics committee and data system do different things. A translation that transfers power to another actor can be wrong without altering any technical noun. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 169: A validation warning identifies an out-of-range entry.

Controlled source excerpt (fictional): “A validation warning identifies an out-of-range entry.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The warning says the participant is medically unsafe.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The message describes the data rule without introducing a clinical judgement. Software validation is not diagnosis. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Test error messages. Second lens — the scope-and-exception check: Underline every qualifier: if, unless, only, at least, no later than, optional, possible, expected, in the named group or under the stated circumstances. Test whether the target sentence covers exactly the same cases. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 170: A form displays a coded visit identifier.

Controlled source excerpt (fictional): “A form displays a coded visit identifier.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The translator replaces it with a descriptive name that conflicts with the schedule.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Preserve the visit ID and localise only the permitted label. Code-label mismatch can attach data to the wrong event. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Compare visit mapping. Second lens — the source-version check: Record the document identifier and revision. Match the line to the currently authorised study document; do not repair the scientific source by intuition. If the source conflicts with another controlled file, open a clarification ticket. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Trial roles, oversight and accountability

Trials use precise roles: participant, legally authorised representative, investigator, sponsor, monitor, ethics committee, data manager and independent reviewer. Translation must not move authority from one role to another.

Diagnostic 171: The principal investigator is responsible for specified site duties.

Controlled source excerpt (fictional): “The principal investigator is responsible for specified site duties.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The translator is the responsible investigator.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The principal investigator retains the named site responsibilities. A linguistic contributor does not inherit clinical accountability. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check role and signature blocks. Second lens — the numeric-and-structured-data check: Compare figures, units, groups, time windows, codes, ordered answers, signatures and references separately from prose. The correct target-language sentence can still be dangerous if an adjacent field contains a copied numerical error. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 172: The sponsor is responsible for the stated trial processes.

Controlled source excerpt (fictional): “The sponsor is responsible for the stated trial processes.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The funding donor individually approves every clinical decision.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The sponsor role follows the source’s specified responsibilities. Sponsor and donor may not be the same entity or role. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Verify definitions in the study. Second lens — the comprehension check: Ask the intended audience to explain what they understand without prompting. They should be able to describe the relevant choice or evidence in ordinary words. Misunderstanding is diagnostic feedback, not a reason to simplify away a right or a risk. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 173: The ethics committee reviewed the consent material.

Controlled source excerpt (fictional): “The ethics committee reviewed the consent material.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The committee personally treats each participant.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The ethics body performs the identified review function. Oversight is not clinical treatment. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Keep functions and participants separate. Second lens — the escalation check: Some ambiguity is not translatable until the source owner answers it. Record the question, candidate meanings, clinical impact, responsible reviewer and authorised resolution. A silent guess is not quality assurance. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 174: The data monitoring committee reviewed interim evidence.

Controlled source excerpt (fictional): “The data monitoring committee reviewed interim evidence.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The committee guaranteed the future result.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The independent group reviewed specified data and made the stated recommendation. Review is not proof of outcome. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Preserve recommendation versus decision. Second lens — the cross-document check: Look for the same concept in the information sheet, protocol, schedule, form, registry and results text. If one version describes a different action or population, establish whether the source documents legitimately differ or whether the translation has drifted. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 175: The monitor checks compliance and records findings.

Controlled source excerpt (fictional): “The monitor checks compliance and records findings.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The monitor is automatically the treating physician.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: A monitor carries out the stated oversight tasks. Clinical care and monitoring are distinct roles. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Review delegation. Second lens — the reader-decision check: Picture a person who never sees the original. Would this wording cause them to consent, decline, report, stop, sign, attend, share information or expect a benefit on a different basis? If so, the divergence is substantive even when grammar is flawless. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 176: A legally authorised representative may consent where allowed.

Controlled source excerpt (fictional): “A legally authorised representative may consent where allowed.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Any relative can consent automatically.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The specified legally authorised representative acts under applicable rules. Legal capacity cannot be presumed from family relation. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check actual rules for the real site. Second lens — the evidence-status check: Identify whether the phrase reports a plan, a possibility, an observation, a provisional assessment, an approved decision or a final result. Preserve that evidence level. Do not use later information to make the source sound more certain. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 177: The investigator explains the research procedures.

Controlled source excerpt (fictional): “The investigator explains the research procedures.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The marketing office is the only source of medical clarification.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Authorised study personnel explain the specified procedures. Accountability belongs to named qualified roles. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Keep contact routes correct. Second lens — the role-and-authority check: Identify the actor named in the controlled record. A study nurse, participant, investigator, sponsor, ethics committee and data system do different things. A translation that transfers power to another actor can be wrong without altering any technical noun. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 178: An independent translator reviews the language.

Controlled source excerpt (fictional): “An independent translator reviews the language.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The translator certifies full medical efficacy.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The translator assesses linguistic quality within the specified process. Language verification is not an efficacy finding. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Separate quality certificates from medical approvals. Second lens — the scope-and-exception check: Underline every qualifier: if, unless, only, at least, no later than, optional, possible, expected, in the named group or under the stated circumstances. Test whether the target sentence covers exactly the same cases. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 179: The sponsor approves a document for controlled distribution.

Controlled source excerpt (fictional): “The sponsor approves a document for controlled distribution.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The document automatically has every required external approval.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Internal release approval is not the same as ethics or regulatory authorisation. Approval layers require distinct evidence. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Record review status per authority. Second lens — the source-version check: Record the document identifier and revision. Match the line to the currently authorised study document; do not repair the scientific source by intuition. If the source conflicts with another controlled file, open a clarification ticket. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 180: A safety decision was made by the designated clinician.

Controlled source excerpt (fictional): “A safety decision was made by the designated clinician.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “A machine translation system made the decision.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The responsible qualified role made the stated clinical decision. Tool output must not be assigned professional authority it lacks. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check decision log and sign-off. Second lens — the numeric-and-structured-data check: Compare figures, units, groups, time windows, codes, ordered answers, signatures and references separately from prose. The correct target-language sentence can still be dangerous if an adjacent field contains a copied numerical error. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Complex sentences, negation and logical qualifiers

Clinical language frequently fails at the level of logic. A missing not, a misplaced only or an ambiguous unless can change eligibility, risk and reporting. The translation must retain the truth conditions of the source.

Diagnostic 181: Do not take the listed action unless authorised by the study team.

Controlled source excerpt (fictional): “Do not take the listed action unless authorised by the study team.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “You may take the action whenever you prefer.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The action is prohibited without the stated authorisation. A missing unless clause changes permission. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Mark logical conditions for independent review. Second lens — the comprehension check: Ask the intended audience to explain what they understand without prompting. They should be able to describe the relevant choice or evidence in ordinary words. Misunderstanding is diagnostic feedback, not a reason to simplify away a right or a risk. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 182: Only participants meeting both conditions are eligible.

Controlled source excerpt (fictional): “Only participants meeting both conditions are eligible.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Either condition alone is sufficient.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Both stated conditions must be satisfied. AND and OR cannot be interchanged. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Draw a truth table for ambiguous criteria. Second lens — the cross-document check: Look for the same concept in the information sheet, protocol, schedule, form, registry and results text. If one version describes a different action or population, establish whether the source documents legitimately differ or whether the translation has drifted. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 183: Participants may discontinue product use while continuing follow-up.

Controlled source excerpt (fictional): “Participants may discontinue product use while continuing follow-up.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Participants who stop product must leave every part of the study.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Product discontinuation and follow-up are separate as stated. One verb cannot swallow an unrelated research activity. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Check programme states. Second lens — the release-state check: A finished translation is not automatically approved for use. Verify scientific review, language review, required approvals, version status and distribution scope, following the study’s governance and applicable requirements. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 184: The results did not rule out a possible small effect.

Controlled source excerpt (fictional): “The results did not rule out a possible small effect.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The results proved there was no effect.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: A small effect remains possible under the evidence described. Negation scope alters scientific inference. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Compare interval and narrative. Second lens — the reader-decision check: Picture a person who never sees the original. Would this wording cause them to consent, decline, report, stop, sign, attend, share information or expect a benefit on a different basis? If so, the divergence is substantive even when grammar is flawless. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 185: The procedure is not required for every participant.

Controlled source excerpt (fictional): “The procedure is not required for every participant.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The procedure is never required.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The obligation varies under specified conditions. Not universal does not mean never. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Identify which subgroup is actually affected. Second lens — the evidence-status check: Identify whether the phrase reports a plan, a possibility, an observation, a provisional assessment, an approved decision or a final result. Preserve that evidence level. Do not use later information to make the source sound more certain. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 186: The adverse event was not assessed as related at the time.

Controlled source excerpt (fictional): “The adverse event was not assessed as related at the time.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The event is proven unrelated forever.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The stated assessment at that time did not find it related. An assessment can change with evidence. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Retain date and assessor. Second lens — the role-and-authority check: Identify the actor named in the controlled record. A study nurse, participant, investigator, sponsor, ethics committee and data system do different things. A translation that transfers power to another actor can be wrong without altering any technical noun. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 187: Unless the criterion is met, the visit will not proceed.

Controlled source excerpt (fictional): “Unless the criterion is met, the visit will not proceed.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The visit proceeds precisely when the criterion is absent.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The visit requires the specified condition before proceeding. Inverting unless changes decision logic. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Test both true and false cases. Second lens — the scope-and-exception check: Underline every qualifier: if, unless, only, at least, no later than, optional, possible, expected, in the named group or under the stated circumstances. Test whether the target sentence covers exactly the same cases. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 188: The study does not offer a guaranteed benefit.

Controlled source excerpt (fictional): “The study does not offer a guaranteed benefit.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “The study offers no conceivable benefit to anyone.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: No benefit is guaranteed; potential benefit remains as described elsewhere. Lack of guarantee is not proof of impossibility. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Compare benefits and uncertainty sections. Second lens — the source-version check: Record the document identifier and revision. Match the line to the currently authorised study document; do not repair the scientific source by intuition. If the source conflicts with another controlled file, open a clarification ticket. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 189: The contact must be notified no later than the stated deadline.

Controlled source excerpt (fictional): “The contact must be notified no later than the stated deadline.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “Notification is only required after that deadline.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: The notification must occur by the specified latest time. Deadline direction is decision-critical. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Cross-check time words and dates. Second lens — the comprehension check: Ask the intended audience to explain what they understand without prompting. They should be able to describe the relevant choice or evidence in ordinary words. Misunderstanding is diagnostic feedback, not a reason to simplify away a right or a risk. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Diagnostic 190: The exception applies to participants who meet the stated conditions only.

Controlled source excerpt (fictional): “The exception applies to participants who meet the stated conditions only.” The first task is to read the sentence together with its document title, intended audience and any surrounding definition. It may be a participant-facing statement, a data record or a study instruction; the translator must not silently promote it to another kind of statement. A source excerpt alone is not enough to decide its exact rendering in another language.

Misleading interpretation: “All participants receive the exception.” This is the type of inference a superficially natural but poorly controlled translation might invite. It is not offered as an acceptable translation. The defect lies in what a reader could conclude about permission, timing, evidence, actor, classification, burden or benefit. Those facts, not verbal elegance, determine whether the target survives review.

Meaning-preserving direction: Only the qualifying group receives the specified exception. Scope of only determines who is affected. Compare the original and proposed meaning in two columns, then ask a bilingual reviewer to explain the practical distinction without relying on the source sentence being visible. Where the document’s authority or language conventions require a different expression, use the controlling definition and record the decision in the project’s glossary.

Specific diagnostic: Highlight qualifiers during review. Second lens — the escalation check: Some ambiguity is not translatable until the source owner answers it. Record the question, candidate meanings, clinical impact, responsible reviewer and authorised resolution. A silent guess is not quality assurance. An acceptable result is not merely that the target words sound professional; it is that the same constrained research claim remains available to a reader in a different language. Record the accepted version, the reviewer, the risk level and any unresolved source ambiguity before release.

Transfer beyond the worked examples: what the method changes

1. From words to decisions

When revising a consent document, annotate the reader’s required decision beside each paragraph. An invitation needs to remain an invitation; a risk needs to remain a risk; a right needs to remain a right. This annotation stops good-sounding wording from defeating the document’s purpose. An experienced reviewer can demonstrate the principle with a before-and-after comparison, the controlling source and a short explanation of the decision that would otherwise have been changed. Keep the repair anchored to evidence rather than describing it only as a matter of style.

2. From one document to a controlled family

A protocol, consent sheet, study calendar, questionnaire, safety notice and registry summary should share a common concept map while retaining their different jobs. An inconsistency may reflect a legitimate genre distinction or a real defect; investigate rather than forcing identical sentences everywhere. An experienced reviewer can demonstrate the principle with a before-and-after comparison, the controlling source and a short explanation of the decision that would otherwise have been changed. Keep the repair anchored to evidence rather than describing it only as a matter of style.

3. From glossary to term governance

A useful glossary includes preferred wording, prohibited confusions, definitions, context examples, version, approver and the fields where a term is valid. A term may be acceptable in scientific reporting but inappropriate in participant-facing explanation unless it is clarified. An experienced reviewer can demonstrate the principle with a before-and-after comparison, the controlling source and a short explanation of the decision that would otherwise have been changed. Keep the repair anchored to evidence rather than describing it only as a matter of style.

4. From readability to genuine comprehension

Short sentences help but are not enough. A person may read a simple sentence and still misunderstand randomisation or personal benefit. Check understanding of the decision: what can be declined, what remains unknown, what happens next and whom to contact. An experienced reviewer can demonstrate the principle with a before-and-after comparison, the controlling source and a short explanation of the decision that would otherwise have been changed. Keep the repair anchored to evidence rather than describing it only as a matter of style.

5. From quality as proofreading to quality as risk management

Allocate specialist effort to statements with high consequences. A typo in a heading and a missing negation in an eligibility rule are not equivalent. Establish severity, likelihood, detectability and whether the error could affect participant choice, safety, analysis or approval. An experienced reviewer can demonstrate the principle with a before-and-after comparison, the controlling source and a short explanation of the decision that would otherwise have been changed. Keep the repair anchored to evidence rather than describing it only as a matter of style.

6. From translation memory to source-version memory

Reuse prior accepted language only if the clinical source, audience, regulatory context and study version make that reuse valid. A translation-memory match can be linguistically exact while scientifically wrong after a protocol amendment. An experienced reviewer can demonstrate the principle with a before-and-after comparison, the controlling source and a short explanation of the decision that would otherwise have been changed. Keep the repair anchored to evidence rather than describing it only as a matter of style.

7. From automated help to accountable review

Machine tools can support retrieval, comparisons, term consistency and draft preparation. They do not automatically supply approval, interpret complex clinical ambiguity or accept responsibility for participant comprehension. Keep a named human decision owner for consequential material. An experienced reviewer can demonstrate the principle with a before-and-after comparison, the controlling source and a short explanation of the decision that would otherwise have been changed. Keep the repair anchored to evidence rather than describing it only as a matter of style.

8. From one language to cross-language symmetry

Review language versions against one common source baseline. A translated form should not have weaker risk warnings, different optional boxes or a stronger promise of treatment than a parallel language version. Symmetry concerns rights and claims, not identical sentence lengths. An experienced reviewer can demonstrate the principle with a before-and-after comparison, the controlling source and a short explanation of the decision that would otherwise have been changed. Keep the repair anchored to evidence rather than describing it only as a matter of style.

9. From publication to correction discipline

When a meaningful error is found after release, trace where the version was distributed, who may have relied on it and what controlling approvals are needed for correction. Preserve the audit trail; do not silently rewrite the past. An experienced reviewer can demonstrate the principle with a before-and-after comparison, the controlling source and a short explanation of the decision that would otherwise have been changed. Keep the repair anchored to evidence rather than describing it only as a matter of style.

10. From generalisation to jurisdictional caution

The same principle of comprehensible, voluntary communication appears across many professional frameworks, but consent processes, signatures, witnesses, storage periods and approval routes can differ. Name specific rules when relevant and never copy a local procedure as a universal requirement. An experienced reviewer can demonstrate the principle with a before-and-after comparison, the controlling source and a short explanation of the decision that would otherwise have been changed. Keep the repair anchored to evidence rather than describing it only as a matter of style.

11. From terminology training to professional judgement

A trainee should learn why adverse event is not automatically adverse reaction, why a placebo group is not necessarily denied all care, and why a nonsignificant result is not proof of equivalence. These distinctions transfer across languages because they are conceptual. An experienced reviewer can demonstrate the principle with a before-and-after comparison, the controlling source and a short explanation of the decision that would otherwise have been changed. Keep the repair anchored to evidence rather than describing it only as a matter of style.

12. From the final document to the lived conversation

An approved translation still needs to be encountered by a reader who has enough time, appropriate assistance and access to questions. The quality system should include how the text is discussed and revisited, while leaving formal consent to authorised research staff. An experienced reviewer can demonstrate the principle with a before-and-after comparison, the controlling source and a short explanation of the decision that would otherwise have been changed. Keep the repair anchored to evidence rather than describing it only as a matter of style.

A practical glossary: controlled clinical research distinctions

1. Adverse event

An untoward medical occurrence in the context specified; not automatically proof that the investigational product caused it. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

2. Adverse reaction

An event interpreted under a defined relationship to a product; use the governing reference definitions. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

3. Serious

A regulatory or protocol classification linked to specified consequences or criteria; not simply a synonym for intense. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

4. Severe

An intensity or grade description, often tied to an explicit scale; not automatically a seriousness criterion. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

5. Expectedness

Relationship to the adverse reactions described in reference safety information; depends on the applicable framework. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

6. Causality

A judgement about a possible cause-and-effect relationship, distinct from the timing of an event. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

7. Safety signal

Information that may justify further assessment, not necessarily a confirmed safety conclusion. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

8. Randomisation

Assignment by a random process within a specified design; not a clinician’s preferred treatment choice. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

9. Allocation concealment

Measures preventing foreknowledge of assignments at the relevant time; distinct from later masking. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

10. Blinding or masking

Restriction of specified assignment information from named roles; the exact roles matter. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

11. Placebo

An inactive comparator relative to a stated active intervention; does not describe all other care. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

12. Active comparator

A comparison intervention with active properties, defined in the study design. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

13. Standard of care

The comparator or care context as defined in the trial; do not assume it is universal. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

14. Eligibility

Satisfaction of the study’s approved inclusion and exclusion criteria at the defined assessment. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

15. Screening

Procedures used to assess whether the study criteria are met; distinct from confirmed enrolment. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

16. Enrolment

Entry into a study according to its definitions; not synonymous with every expression of interest. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

17. Treatment arm

A study group defined by allocated intervention or comparator; not necessarily a clinical ward. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

18. Protocol

The controlled research plan, including objectives, design, procedures, monitoring and specified analyses. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

19. Amendment

A controlled change to the study documentation with a defined status and implementation. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

20. Deviation

A departure from approved protocol or process; not automatically the same as an amendment. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

21. Consent process

The communication and decision process associated with legally and ethically valid participation. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

22. Informed consent form

A specified record of consent information and documentation; signatures do not replace understanding. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

23. Assent

A form of agreement used in certain contexts with persons who cannot provide the relevant legal consent; use local and protocol definitions. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

24. Legally authorised representative

A person permitted by applicable rules to act in a defined consent capacity. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

25. Withdrawal

Ending some or all participation under the described rights and conditions; not automatically data erasure. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

26. Reconsent

A renewed consent process when required after relevant new information or change; not every revision automatically mandates it. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

27. Primary endpoint

A prespecified main measure used to evaluate the scientific question under the protocol. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

28. Secondary endpoint

An additional specified outcome, with interpretation dependent on analysis design. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

29. Exploratory endpoint

A research measure often used to generate further hypotheses or understanding. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

30. Surrogate endpoint

A measure intended to stand in for a clinical outcome; validity depends on evidence and context. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

31. Analysis population

The people whose data contribute to a specific analysis under its definitions. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

32. Intention-to-treat

An analysis principle often relating participants to assigned groups; details depend on the planned method. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

33. Per-protocol

An analysis set or method applying protocol-adherence criteria defined by the study. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

34. Denominator

The population or count against which a proportion or rate is calculated. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

35. Confidence interval

An interval estimate with defined statistical interpretation; not a promise of an individual’s result. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

36. Risk ratio

A comparative statistical measure whose reference group and time window must be clear. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

37. Hazard ratio

A time-to-event comparative measure requiring context; not automatically a simple ratio of event probabilities. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

38. Statistical significance

A conclusion under a statistical procedure and threshold; not equivalent to clinical importance. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

39. Clinical importance

Practical relevance for care or outcomes, a question separate from statistical testing. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

40. Patient-reported outcome

An outcome reported by a participant rather than inferred or assessed by another person. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

41. Validated instrument

A measure with evidence supporting particular use and interpretation; wording changes may affect validity. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

42. Recall period

The time window a question asks a respondent to consider. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

43. Coding

Assignment of defined categories or controlled terms to data; not a free translation of machine identifiers. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

44. Pseudonymisation

Use of separated identifiers and safeguards while a re-linking pathway may remain. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

45. Anonymisation

Processing aiming to remove identification under the relevant technical and legal criteria. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

46. Data retention

The controlled period and conditions under which study records are stored. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

47. Essential records

Documents that support evaluation of trial conduct and data reliability in the relevant framework. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

48. Monitoring

Oversight and checking processes distinct from directly treating a participant. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

49. Auditing

A systematic, independent examination under the relevant governance context. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

50. Independent ethics committee

An oversight body reviewing specified ethical aspects according to applicable rules. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

51. Sponsor

The party assigned sponsor responsibilities in the trial framework, not necessarily a generic donor. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

52. Investigator

A professional responsible for the specified conduct of research at a site. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

53. Comparator

The condition or intervention against which another is evaluated in the defined design. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

54. Investigational product

The product being studied or used as specified in the research design. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

55. Concomitant therapy

Other treatment used during the trial, whose status and restrictions require protocol context. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

56. Dose modification

A controlled change in administration under specified clinical and protocol conditions. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

57. Visit window

An authorised range for a study assessment, distinct from a fixed appointment. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

58. Data lock

A defined state in data preparation or analysis, not necessarily the end of all study work. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

59. Protocol-specified analysis

An analysis set out in controlled study plans before the relevant evidence is examined. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

60. Interim analysis

An analysis at an intermediate stage, with permitted interpretation dependent on the study design. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

61. Final report

A completed research report according to the defined documentation process. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

62. Trial registry

A public system recording specified study details; it is not automatically the source of live medical advice. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

63. Plain-language summary

A reader-friendly account that must still preserve material claims, limitations and uncertainty. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

64. Translation revision

Quality review of a translation against the source by a competent person; not the same as source editing. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

65. Translation review

Assessment for suitability, specialist terminology or audience using an agreed quality process. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

66. Interpreting

Transfer of spoken or signed communication between languages during an interaction. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

67. Comprehension check

A method of testing whether key information has been understood, conducted appropriately for the context. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

68. Termbase

A managed collection of term concepts, definitions, contexts, approved labels and usage restrictions. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

69. Source version

The exact controlled document state from which a translation was produced. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

70. Approval status

Recorded authorisation by a named party for a specified version and use. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

71. Audit trail

A record of actions, changes, actors and times supporting traceability. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

72. Residual uncertainty

Unresolved doubt remaining after review, requiring a recorded decision rather than a concealed guess. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

73. Controlled vocabulary

An approved set of labels and codes with specified mappings and meanings. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

74. Scope qualifier

Language specifying when, for whom or under what conditions a statement applies. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

75. Negative finding

A result that may refer to absence of evidence or specific measured absence; not universally equivalent to zero. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

76. Reporting date

Date information was submitted or received, different from occurrence or onset date. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

77. Participant-facing material

Content intended for a person’s understanding and decisions, not merely investigator documentation. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

78. Conditional statement

A claim whose truth or required action depends on a specified condition. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

79. Unblinding

Disclosure of assignment under defined circumstances and authority. For translation, verify the term in its full clause: the speaker, object, qualifier, timeline and relevant protocol definition determine which target-language expression is defensible. A useful glossary records a good contextual example and at least one false friend or misleading alternative where such an error is plausible.

Common questions about clinical trial translation

What is clinical trial translation?

It is the controlled transfer of research documents and communications between languages without changing research meaning, participant choices or recorded evidence. A practical test is to compare the answer a target-language reader would give against the current controlling source, not against a convenient older template. If the query concerns a participant’s own situation, direct it to the authorised research team rather than using an educational article to make an individual medical or legal determination.

Does a translated consent form automatically make consent valid?

No. Consent is a process, and its legal and ethical requirements vary by jurisdiction and study. Accurate language is important but cannot replace a lawful, understandable discussion and applicable approvals. A practical test is to compare the answer a target-language reader would give against the current controlling source, not against a convenient older template. If the query concerns a participant’s own situation, direct it to the authorised research team rather than using an educational article to make an individual medical or legal determination.

Is a certified translator always legally required?

Requirements vary. The needed competence, declarations and approvals depend on the document, sponsoring institution, ethics body and governing law. Do not make a universal claim. A practical test is to compare the answer a target-language reader would give against the current controlling source, not against a convenient older template. If the query concerns a participant’s own situation, direct it to the authorised research team rather than using an educational article to make an individual medical or legal determination.

What is the difference between medical translation and clinical research translation?

Both may use medical language, but research translation adds design, allocation, trial status, consent, protocol versioning, adverse-event classifications and evidence reporting. A practical test is to compare the answer a target-language reader would give against the current controlling source, not against a convenient older template. If the query concerns a participant’s own situation, direct it to the authorised research team rather than using an educational article to make an individual medical or legal determination.

Can machine translation be used?

Tools may assist parts of a workflow, subject to privacy, quality and governance controls. Raw output is not an approval and is particularly risky for rights, safety and technical classifications. A practical test is to compare the answer a target-language reader would give against the current controlling source, not against a convenient older template. If the query concerns a participant’s own situation, direct it to the authorised research team rather than using an educational article to make an individual medical or legal determination.

What is an informed consent form?

It is one document supporting the consent process. It usually describes material aspects of research, but what must be included and how consent is obtained depend on governing requirements. A practical test is to compare the answer a target-language reader would give against the current controlling source, not against a convenient older template. If the query concerns a participant’s own situation, direct it to the authorised research team rather than using an educational article to make an individual medical or legal determination.

Why not simply translate every technical word literally?

A literal rendering may convey the wrong concept. The target should preserve the source’s research distinction while using language comprehensible to the intended audience. A practical test is to compare the answer a target-language reader would give against the current controlling source, not against a convenient older template. If the query concerns a participant’s own situation, direct it to the authorised research team rather than using an educational article to make an individual medical or legal determination.

What is a back translation?

A separate rendering of the translated text back toward the source language. It can identify discrepancies but is not by itself proof of conceptual equivalence or participant comprehension. A practical test is to compare the answer a target-language reader would give against the current controlling source, not against a convenient older template. If the query concerns a participant’s own situation, direct it to the authorised research team rather than using an educational article to make an individual medical or legal determination.

How do I translate randomisation?

Explain that assignment uses a chance mechanism as described, keep the ratio and group names, and avoid implying that clinician or participant chooses the group. A practical test is to compare the answer a target-language reader would give against the current controlling source, not against a convenient older template. If the query concerns a participant’s own situation, direct it to the authorised research team rather than using an educational article to make an individual medical or legal determination.

How should placebo be explained?

Use the protocol’s exact comparator description and its surrounding treatment context. Do not claim a placebo group receives no care unless the source actually supports that statement. A practical test is to compare the answer a target-language reader would give against the current controlling source, not against a convenient older template. If the query concerns a participant’s own situation, direct it to the authorised research team rather than using an educational article to make an individual medical or legal determination.

Is an adverse event always caused by the study medicine?

No. The reporting of an event and the assessment of a causal relationship are different matters. A practical test is to compare the answer a target-language reader would give against the current controlling source, not against a convenient older template. If the query concerns a participant’s own situation, direct it to the authorised research team rather than using an educational article to make an individual medical or legal determination.

Are serious and severe synonyms?

No. Seriousness may follow defined reporting criteria; severity describes intensity or grading. Translate the classification actually present. A practical test is to compare the answer a target-language reader would give against the current controlling source, not against a convenient older template. If the query concerns a participant’s own situation, direct it to the authorised research team rather than using an educational article to make an individual medical or legal determination.

Can a participant withdraw and erase all their data?

The right to withdraw participation and the treatment of previously collected data depend on the approved information and applicable requirements. A translation must not promise universal erasure. A practical test is to compare the answer a target-language reader would give against the current controlling source, not against a convenient older template. If the query concerns a participant’s own situation, direct it to the authorised research team rather than using an educational article to make an individual medical or legal determination.

How should dates and doses be checked?

Compare them independently against controlled values, units, routes, time windows and formats; escalate any ambiguity to qualified personnel. A practical test is to compare the answer a target-language reader would give against the current controlling source, not against a convenient older template. If the query concerns a participant’s own situation, direct it to the authorised research team rather than using an educational article to make an individual medical or legal determination.

What happens when a protocol is amended?

Relevant translations may need changes and renewed review. Responsible parties determine which approvals, notices and consent actions are required. A practical test is to compare the answer a target-language reader would give against the current controlling source, not against a convenient older template. If the query concerns a participant’s own situation, direct it to the authorised research team rather than using an educational article to make an individual medical or legal determination.

How do you translate trial results fairly?

Keep the prespecified question, comparison, analysis population, effect size, interval, limitations and distinction between statistical and clinical importance. A practical test is to compare the answer a target-language reader would give against the current controlling source, not against a convenient older template. If the query concerns a participant’s own situation, direct it to the authorised research team rather than using an educational article to make an individual medical or legal determination.

What is a plain-language summary?

A reader-oriented explanation of the research and its findings. It should make evidence accessible without hiding uncertainty or implying proof beyond the results. A practical test is to compare the answer a target-language reader would give against the current controlling source, not against a convenient older template. If the query concerns a participant’s own situation, direct it to the authorised research team rather than using an educational article to make an individual medical or legal determination.

Can a glossary replace specialist review?

No. It helps maintain consistency but cannot resolve an ambiguous protocol or determine a clinical conclusion. A practical test is to compare the answer a target-language reader would give against the current controlling source, not against a convenient older template. If the query concerns a participant’s own situation, direct it to the authorised research team rather than using an educational article to make an individual medical or legal determination.

What should a study translation brief contain?

Document IDs and versions, audience, source authority, field definitions, terminology, technical restrictions, review roles, deadlines and escalation contacts. A practical test is to compare the answer a target-language reader would give against the current controlling source, not against a convenient older template. If the query concerns a participant’s own situation, direct it to the authorised research team rather than using an educational article to make an individual medical or legal determination.

How are multilingual questionnaires checked?

Review intended constructs, recall periods, anchors, code mappings, branching, licensing and validated versions where required. A practical test is to compare the answer a target-language reader would give against the current controlling source, not against a convenient older template. If the query concerns a participant’s own situation, direct it to the authorised research team rather than using an educational article to make an individual medical or legal determination.

Does eConsent change translation obligations?

It adds interface and data-flow checks. Wording, button logic, accessibility, displayed version and signature records must remain aligned. A practical test is to compare the answer a target-language reader would give against the current controlling source, not against a convenient older template. If the query concerns a participant’s own situation, direct it to the authorised research team rather than using an educational article to make an individual medical or legal determination.

Can consent rules from one country be applied everywhere?

No. Use international principles for orientation and verify current requirements for each actual trial location and approval pathway. A practical test is to compare the answer a target-language reader would give against the current controlling source, not against a convenient older template. If the query concerns a participant’s own situation, direct it to the authorised research team rather than using an educational article to make an individual medical or legal determination.

What is a safe first step when the source is unclear?

Record the ambiguity and ask the responsible study or protocol owner to clarify it in the controlled source. Do not guess. A practical test is to compare the answer a target-language reader would give against the current controlling source, not against a convenient older template. If the query concerns a participant’s own situation, direct it to the authorised research team rather than using an educational article to make an individual medical or legal determination.

What is the final approval gate?

Documented release through the specific study governance process, including qualified language and scientific review and any applicable ethics or regulatory approval. A practical test is to compare the answer a target-language reader would give against the current controlling source, not against a convenient older template. If the query concerns a participant’s own situation, direct it to the authorised research team rather than using an educational article to make an individual medical or legal determination.

Next routes: translate meaning, learn the words, understand the language

This specialist volume belongs beneath Master Art of Translation | The Complete System for Moving Meaning Between Languages. Go there for general translation theory, the source–meaning–target method, terminology control, quality and transfer across unrelated disciplines. The article here deliberately owns clinical trial and consent communication only; it does not replace the translation root or other specialist children.

If the difficulty is learning the words rather than controlling a trial document, use the protected Vocabulary Learning Hub. Its role is vocabulary learning and progression, not an alternative source of clinical research standards. The distinction is useful: learning the difference between possible and probable belongs to vocabulary; deciding which qualifier an approved protocol uses belongs to its controlled clinical translation process.

For sentence meaning, clause relationships, hedging, explanation, audience and careful paraphrase, continue to the protected How English Works | Translation and Meaning Across Languages. Return to this guide when that knowledge must be applied to trial protocols, participant rights, safety categories and evidence. No protected learning owner is rewritten by this new volume.

High-authority sources, dates and scope

International Council for Harmonisation (ICH), official site — Current ICH guideline programme, GCP E6(R3), multidisciplinary standards and MedDRA context; check regional adoption.

European Medicines Agency, ICH E6 Good Clinical Practice — Principles and Annex 1 effective in the EU from 23 July 2025; Annex 2 scheduled to apply there on 15 January 2027, per EMA status checked 30 October 2026.

U.S. FDA, Informed Consent Guidance (August 2023) — Official guidance on the roles of IRBs, investigators and sponsors; a United States reference, not a universal global rule.

U.S. FDA, 21 CFR Part 50 informed consent requirements — A concrete jurisdictional example of understandable consent language and the distinction between law and educational best practice.

World Health Organization, ICTRP data and transparency — Trial registration transparency, data ecosystem and the importance of correctly identifying registry information.

World Health Organization, ICTRP search portal — WHO notes that search-portal datasets are in English, while individual source registries may offer additional languages.

ISO, ISO 17100:2015 translation services requirements — Quality process and professional-resource framework; ISO notes raw machine translation plus post-editing and interpreting are outside its scope.

Johns Hopkins Medicine IRB, consent for non-English speakers (May 2026) — Institutional example of translated written materials and interpreting for informed consent; site-specific procedure rather than global law.

United States federal human-subject regulation, 45 CFR 46.116 — A jurisdictional example requiring understandable language and a presentation of key information that supports comprehension.

Research currency: Sources checked 30 October 2026. Regulatory implementation varies across jurisdictions, including the timing of revised GCP provisions. The 2026 consolidated ICH E6(R3) material and scheduled 2027 Annex 2 application in the EU must not be described as universally effective on the same date. Verify controlling editions and local requirements for actual work.

Scope: This is a world-facing educational translation reference using fictional diagnostic examples. It does not constitute legal advice, clinical advice, a participant information sheet, a research protocol, verified target-language consent text, an approved ethics submission or a regulatory compliance certificate. Current source records, qualified investigators, ethics bodies and relevant authorities govern real studies.