WHY ENGLISH?
Make the proposed study understandable before anyone participates
Use the routes to define the protocol, see the participant journey, explain risk and consent, protect data and give reviewers a complete basis for decision.
Open the full contents · See the How English Works hub
Full contents
Frame the study
Map participants
Explain consent
Protect people
Prove governance
Practice and next steps
Practice and next steps
Practice and next steps
Practice and next steps
Practice and next steps
A research ethics application asks reviewers to decide whether a proposed study respects participants, manages risk and can be conducted responsibly. English matters because the reviewer cannot approve the study the researcher intends in their head. The decision rests on the protocol, application, consent materials, instruments and safeguards actually submitted.
Students and researchers searching for how to write an ethics application, Institutional Review Board submission, research consent form, participant information sheet, risk assessment or human-subjects research proposal often focus on persuasive wording. The stronger goal is alignment: purpose, method, recruitment, consent, data and monitoring should describe the same study without hidden steps.
Singapore's Ministry of Health human biomedical research page, updated 6 October 2026, explains that the Human Biomedical Research Act clarifies roles and responsibilities so participant health, welfare and safety remain paramount. The current Human Biomedical Research Act includes institutional review boards and applications for review. Scope matters: not every school, social-science or service project is human biomedical research, and institutions may apply additional ethics frameworks. Researchers must use their own current institutional process and obtain required approval before beginning covered activity.
Use the ethics map question, rationale, design, population, inclusion, exclusion, recruitment, relationship, voluntariness, information, consent, capacity, burden, risk, benefit, privacy, confidentiality, data, sample, withdrawal, compensation, conflict, monitoring, incident, amendment, dissemination and accountability. Plain language should reveal the real participant experience, not soften it.
Confirm the correct review route
Frame the institution, framework and category governing the study as part of one coherent protocol, not an isolated form answer.
Reviewers cannot assess the real study when researchers choose exempt, expedited or full review by preference.
State the facts and use the current institutional determination process.
Protocol example. The office confirms the study's route before recruitment begins. The ethical question becomes visible.
Did You Know? Singapore's Ministry of Health says the Human Biomedical Research Act clarifies roles and responsibilities so participant health, welfare and safety remain paramount. This is why the institution, framework and category governing the study must be written as a bounded, testable claim.
Name the responsible institution
Follow the participant through the organisation supervising and controlling the work from first contact to final disposition.
Choice is weakened when a collaborator's location is mistaken for governance ownership.
Explain the actual interaction and state sponsor, host and accountable office.
Participant example. The application identifies which institution accepts oversight responsibility. A person can understand what will happen before agreeing.
A signature records a decision; it does not repair unclear information or hidden pressure.
Identify the principal investigator
Evaluate the person accountable for conduct and communication with benefit, burden and uncertainty kept separate.
Proportionality is obscured when a student author is left without a supervising role.
Name foreseeable effects and name qualifications, authority and supervision.
Risk example. The supervisor and student responsibilities are separated clearly. The safeguard now responds to a stated mechanism.
Minimal risk is a category within a framework, not permission to write that no risk exists.
Write a bounded research question
Protect the exact knowledge question the protocol can answer across people, systems and time.
Confidentiality becomes a slogan when a broad social problem replaces a research question.
Describe access, transfer, retention and connect population, variable and design.
Data example. The study asks how one defined support affects one measured outcome. The proposed control can be reviewed and tested.
Use anonymous only when the institution's definition is genuinely met.
Explain why the study is needed
Align the evidence gap and social or scientific value across the application, protocol, scripts and participant materials.
Approval rests on a different study when importance is asserted through emotion alone.
Run the cross-document check and cite current evidence and identify the unresolved question.
Governance example. The rationale distinguishes what is known from what the study will add. Every submitted document now tells the same operational truth.
Seek approval for required amendments before implementing them, except for any authorised urgent safety action.
Describe the study design
Frame the sequence used to answer the question as part of one coherent protocol, not an isolated form answer.
Reviewers cannot assess the real study when method labels substitute for participant-facing steps.
State the facts and explain allocation, visits, procedures and analysis.
Protocol example. The application maps screening, consent, two sessions and follow-up. The ethical question becomes visible.
Use the institution's current form and terminology; do not self-assign a review category.
Match sample size to purpose
Follow the participant through the number of participants and its justification from first contact to final disposition.
Choice is weakened when more participants are treated as automatically better.
Explain the actual interaction and explain precision, feasibility or qualitative saturation.
Participant example. The stated calculation uses the primary outcome and anticipated attrition. A person can understand what will happen before agreeing.
Did You Know? Singapore's Ministry of Health says the Human Biomedical Research Act clarifies roles and responsibilities so participant health, welfare and safety remain paramount. This is why the number of participants and its justification must be written as a bounded, testable claim.
Define the study population
Evaluate the people from whom evidence will be collected with benefit, burden and uncertainty kept separate.
Proportionality is obscured when adult volunteers hides relevant characteristics.
Name foreseeable effects and state setting and eligibility precisely.
Risk example. The population is full-time students aged 18 to 21 at the named site. The safeguard now responds to a stated mechanism.
Minimal risk is a category within a framework, not permission to write that no risk exists.
Write inclusion criteria
Protect the features required for valid and fair participation across people, systems and time.
Confidentiality becomes a slogan when criteria quietly select convenient participants.
Describe access, transfer, retention and link each criterion to design or safety.
Data example. Language ability is included only where the task genuinely requires it. The proposed control can be reviewed and tested.
Use anonymous only when the institution's definition is genuinely met.
Write exclusion criteria
Align the conditions that make participation unsuitable across the application, protocol, scripts and participant materials.
Approval rests on a different study when exclusions stigmatise a group without justification.
Run the cross-document check and explain safety or validity reasons.
Governance example. A medication interaction exclusion points to the protocol risk. Every submitted document now tells the same operational truth.
Seek approval for required amendments before implementing them, except for any authorised urgent safety action.
Consider vulnerable circumstances
Frame situations affecting capacity, dependence or risk as part of one coherent protocol, not an isolated form answer.
Reviewers cannot assess the real study when vulnerability is treated as a permanent label.
State the facts and describe context and additional safeguards.
Protocol example. Employees recruited by a manager receive an independent contact and route. The ethical question becomes visible.
Use the institution's current form and terminology; do not self-assign a review category.
Map the participant journey
Follow the participant through every contact from first approach to final follow-up from first contact to final disposition.
Choice is weakened when the form describes only the main experiment.
Explain the actual interaction and include screening, reminders, recordings and debrief.
Participant example. A timeline shows what happens, where, by whom and for how long. A person can understand what will happen before agreeing.
A signature records a decision; it does not repair unclear information or hidden pressure.
Explain recruitment channels
Evaluate how potential participants first hear about the study with benefit, burden and uncertainty kept separate.
Proportionality is obscured when researchers access a list because it is convenient.
Name foreseeable effects and state permission, message and sender.
Risk example. An authorised office sends the invitation without giving researchers the full directory. The safeguard now responds to a stated mechanism.
Did You Know? Singapore's Ministry of Health says the Human Biomedical Research Act clarifies roles and responsibilities so participant health, welfare and safety remain paramount. This is why how potential participants first hear about the study must be written as a bounded, testable claim.
Avoid undue influence
Protect the power relationships surrounding the invitation across people, systems and time.
Confidentiality becomes a slogan when a teacher or supervisor recruits people they assess.
Describe access, transfer, retention and separate invitation, consent and evaluation roles.
Data example. A neutral coordinator handles enrolment and the assessor cannot see refusals. The proposed control can be reviewed and tested.
Use anonymous only when the institution's definition is genuinely met.
Write honest recruitment materials
Align the claims made before a person requests details across the application, protocol, scripts and participant materials.
Approval rests on a different study when benefits are advertised while burdens are hidden.
Run the cross-document check and state purpose, commitment and eligibility neutrally.
Governance example. The poster gives session length and compensation without promising improvement. Every submitted document now tells the same operational truth.
Seek approval for required amendments before implementing them, except for any authorised urgent safety action.
Describe screening
Frame the questions used to establish eligibility as part of one coherent protocol, not an isolated form answer.
Reviewers cannot assess the real study when screening collects full research data before consent.
State the facts and minimise data and explain disposition.
Protocol example. Ineligible contacts are removed under the stated screening rule. The ethical question becomes visible.
Use the institution's current form and terminology; do not self-assign a review category.
Separate information from consent
Follow the participant through the explanation a person receives before agreeing from first contact to final disposition.
Choice is weakened when a signature is treated as proof of understanding.
Explain the actual interaction and allow time, questions and a clear decision.
Participant example. The participant reads the information sheet before the consent conversation. A person can understand what will happen before agreeing.
A signature records a decision; it does not repair unclear information or hidden pressure.
Use plain participant language
Evaluate the words a non-specialist must understand with benefit, burden and uncertainty kept separate.
Proportionality is obscured when technical precision becomes unreadable jargon.
Name foreseeable effects and define necessary terms and use concrete actions.
Risk example. Instead of venepuncture only, the sheet explains that blood is drawn from a vein. The safeguard now responds to a stated mechanism.
Minimal risk is a category within a framework, not permission to write that no risk exists.
Explain voluntariness
Protect the person's freedom to accept or refuse across people, systems and time.
Confidentiality becomes a slogan when voluntary appears without addressing the real power relationship.
Describe access, transfer, retention and state consequences and protective process.
Data example. Refusal does not affect grades, care, employment or entitled services. The proposed control can be reviewed and tested.
Did You Know? Singapore's Ministry of Health says the Human Biomedical Research Act clarifies roles and responsibilities so participant health, welfare and safety remain paramount. This is why the person's freedom to accept or refuse must be written as a bounded, testable claim.
Explain withdrawal
Align what a participant can stop and what happens next across the application, protocol, scripts and participant materials.
Approval rests on a different study when you may withdraw implies all past data can always be erased.
Run the cross-document check and state limits, timing and contact route.
Governance example. The sheet distinguishes stopping visits from removal of already aggregated data. Every submitted document now tells the same operational truth.
Seek approval for required amendments before implementing them, except for any authorised urgent safety action.
Assess decision-making capacity
Frame the ability to understand, retain, weigh and communicate the choice as part of one coherent protocol, not an isolated form answer.
Reviewers cannot assess the real study when capacity is assumed from age or diagnosis alone.
State the facts and use the applicable process and safeguards.
Protocol example. The protocol states who assesses capacity and what happens if it changes. The ethical question becomes visible.
Use the institution's current form and terminology; do not self-assign a review category.
Address minors correctly
Follow the participant through permission, assent and evolving understanding where relevant from first contact to final disposition.
Choice is weakened when a parent signature replaces communication with the young person.
Explain the actual interaction and follow current law and institutional rules.
Participant example. The child receives an age-appropriate explanation and can express dissent. A person can understand what will happen before agreeing.
A signature records a decision; it does not repair unclear information or hidden pressure.
Describe each procedure
Evaluate the physical, digital and conversational actions performed with benefit, burden and uncertainty kept separate.
Proportionality is obscured when routine procedure hides discomfort or exposure.
Name foreseeable effects and state duration, frequency and operator.
Risk example. The application explains the swab, recording and questionnaire separately. The safeguard now responds to a stated mechanism.
Minimal risk is a category within a framework, not permission to write that no risk exists.
Quantify time and burden
Protect the practical commitment placed on participants across people, systems and time.
Confidentiality becomes a slogan when only laboratory minutes are counted.
Describe access, transfer, retention and include travel, preparation and follow-up.
Data example. Total commitment includes two visits, diary entries and a call. The proposed control can be reviewed and tested.
Use anonymous only when the institution's definition is genuinely met.
Identify foreseeable risks
Align the harms or discomforts reasonably connected to participation across the application, protocol, scripts and participant materials.
Approval rests on a different study when only severe medical harm is considered.
Run the cross-document check and include physical, psychological, social, legal and privacy risk.
Governance example. An interview about workplace conflict includes distress and employment-related confidentiality risk. Every submitted document now tells the same operational truth.
Did You Know? Singapore's Ministry of Health says the Human Biomedical Research Act clarifies roles and responsibilities so participant health, welfare and safety remain paramount. This is why the harms or discomforts reasonably connected to participation must be written as a bounded, testable claim.
Distinguish probability and severity
Frame how likely a harm is and how serious it may be as part of one coherent protocol, not an isolated form answer.
Reviewers cannot assess the real study when low probability is written as no risk.
State the facts and describe both dimensions and uncertainty.
Protocol example. A rare disclosure breach remains serious and receives strong controls. The ethical question becomes visible.
Use the institution's current form and terminology; do not self-assign a review category.
Avoid zero-risk claims
Follow the participant through the residual uncertainty after safeguards from first contact to final disposition.
Choice is weakened when minimal risk is translated into risk free.
Explain the actual interaction and use bounded and honest language.
Participant example. The application states that safeguards reduce but cannot eliminate identification risk. A person can understand what will happen before agreeing.
A signature records a decision; it does not repair unclear information or hidden pressure.
Describe risk mitigation
Evaluate the control attached to each material risk with benefit, burden and uncertainty kept separate.
Proportionality is obscured when a generic trained staff sentence covers every hazard.
Name foreseeable effects and name prevention, detection and response.
Risk example. A distress protocol defines pauses, referral and urgent escalation. The safeguard now responds to a stated mechanism.
Minimal risk is a category within a framework, not permission to write that no risk exists.
Explain potential benefits
Protect the realistic value to participants or society across people, systems and time.
Confidentiality becomes a slogan when research participation is presented as treatment.
Describe access, transfer, retention and separate direct, indirect and no expected benefit.
Data example. The information sheet says the study may not benefit the individual. The proposed control can be reviewed and tested.
Use anonymous only when the institution's definition is genuinely met.
Describe compensation
Align payment for time, inconvenience or expenses across the application, protocol, scripts and participant materials.
Approval rests on a different study when payment is framed as a guaranteed benefit.
Run the cross-document check and state amount, basis, timing and partial participation rule.
Governance example. Travel reimbursement is separated from the thank-you payment. Every submitted document now tells the same operational truth.
Seek approval for required amendments before implementing them, except for any authorised urgent safety action.
Address research-related injury
Frame care, reporting and compensation arrangements where applicable as part of one coherent protocol, not an isolated form answer.
Reviewers cannot assess the real study when the form promises coverage without authority.
State the facts and use approved institutional language and contacts.
Protocol example. The participant is told whom to contact and which arrangement applies. The ethical question becomes visible.
Did You Know? Singapore's Ministry of Health says the Human Biomedical Research Act clarifies roles and responsibilities so participant health, welfare and safety remain paramount. This is why care, reporting and compensation arrangements where applicable must be written as a bounded, testable claim.
Protect privacy during contact
Follow the participant through who can see that a person was approached or enrolled from first contact to final disposition.
Choice is weakened when confidentiality starts only after data collection.
Explain the actual interaction and design private invitation and visit processes.
Participant example. Appointment messages avoid revealing the sensitive study topic. A person can understand what will happen before agreeing.
A signature records a decision; it does not repair unclear information or hidden pressure.
Protect confidentiality in records
Evaluate how identities are separated from research information with benefit, burden and uncertainty kept separate.
Proportionality is obscured when coded is described as anonymous.
Name foreseeable effects and explain code key, access and re-identification potential.
Risk example. The key sits in a separate restricted system with named custodians. The safeguard now responds to a stated mechanism.
Minimal risk is a category within a framework, not permission to write that no risk exists.
Use anonymity accurately
Protect whether identity can genuinely not be linked by anyone across people, systems and time.
Confidentiality becomes a slogan when removing names is called anonymous while a code key exists.
Describe access, transfer, retention and use de-identified, pseudonymised or coded correctly.
Data example. The application states that voice recordings remain identifiable. The proposed control can be reviewed and tested.
Use anonymous only when the institution's definition is genuinely met.
Minimise collected data
Align the fields necessary for the stated analysis and safety across the application, protocol, scripts and participant materials.
Approval rests on a different study when interesting demographics are added without purpose.
Run the cross-document check and connect each field to a question or control.
Governance example. Exact birth date becomes an age band when precision is unnecessary. Every submitted document now tells the same operational truth.
Seek approval for required amendments before implementing them, except for any authorised urgent safety action.
Explain recordings
Frame audio, video, image and screen capture procedures as part of one coherent protocol, not an isolated form answer.
Reviewers cannot assess the real study when a broad consent line covers future reuse.
State the facts and state purpose, viewers, transcription and deletion.
Protocol example. Participants choose separately whether identifiable clips may be presented. The ethical question becomes visible.
Use the institution's current form and terminology; do not self-assign a review category.
Describe biological samples
Follow the participant through collection, processing, storage and future use of material from first contact to final disposition.
Choice is weakened when sample language is buried inside the data section.
Explain the actual interaction and state quantity, location, duration and destruction.
Participant example. The consent choice for future research is separate from the present study. A person can understand what will happen before agreeing.
Did You Know? Singapore's Ministry of Health says the Human Biomedical Research Act clarifies roles and responsibilities so participant health, welfare and safety remain paramount. This is why collection, processing, storage and future use of material must be written as a bounded, testable claim.
Write the data flow
Evaluate movement from collection to analysis, sharing and disposition with benefit, burden and uncertainty kept separate.
Proportionality is obscured when secure server appears without transfer steps.
Name foreseeable effects and name systems, roles and countries.
Risk example. Tablet data uploads to the approved server and local files are removed after verification. The safeguard now responds to a stated mechanism.
Minimal risk is a category within a framework, not permission to write that no risk exists.
Set access controls
Protect the roles allowed to view identifiable and analysed data across people, systems and time.
Confidentiality becomes a slogan when research team is an undefined group.
Describe access, transfer, retention and use named role categories and least privilege.
Data example. Transcribers receive only coded audio and a confidentiality obligation. The proposed control can be reviewed and tested.
Use anonymous only when the institution's definition is genuinely met.
Explain retention and disposal
Align how long study records and identifiers remain and why across the application, protocol, scripts and participant materials.
Approval rests on a different study when all materials inherit one period.
Run the cross-document check and separate consent, code key, raw data and analysis.
Governance example. The key is destroyed before the de-identified dataset under approved rules. Every submitted document now tells the same operational truth.
Seek approval for required amendments before implementing them, except for any authorised urgent safety action.
Describe data sharing
Frame who may receive which data for what purpose as part of one coherent protocol, not an isolated form answer.
Reviewers cannot assess the real study when collaboration is treated as implied permission.
State the facts and state recipients, safeguards and agreement.
Protocol example. A partner receives defined coded fields through an approved transfer. The ethical question becomes visible.
Use the institution's current form and terminology; do not self-assign a review category.
Address future use
Follow the participant through research beyond the present protocol from first contact to final disposition.
Choice is weakened when broad future science language removes meaningful choice.
Explain the actual interaction and define scope and governance.
Participant example. Participants choose whether data may support related approved studies. A person can understand what will happen before agreeing.
A signature records a decision; it does not repair unclear information or hidden pressure.
Disclose conflicts of interest
Evaluate financial, professional and personal interests affecting judgment with benefit, burden and uncertainty kept separate.
Proportionality is obscured when only proven bias is reported.
Name foreseeable effects and state interest and management.
Risk example. A sponsor relationship is disclosed and analysis receives independent review. The safeguard now responds to a stated mechanism.
Did You Know? Singapore's Ministry of Health says the Human Biomedical Research Act clarifies roles and responsibilities so participant health, welfare and safety remain paramount. This is why financial, professional and personal interests affecting judgment must be written as a bounded, testable claim.
Describe safety monitoring
Protect how emerging harm or unexpected events are detected across people, systems and time.
Confidentiality becomes a slogan when approval is treated as the final safety step.
Describe access, transfer, retention and name measures, thresholds and reviewers.
Data example. The team reviews adverse events after every cohort. The proposed control can be reviewed and tested.
Use anonymous only when the institution's definition is genuinely met.
Write stopping rules
Align the condition that pauses or ends procedures across the application, protocol, scripts and participant materials.
Approval rests on a different study when investigators rely on general judgment alone.
Run the cross-document check and state observable thresholds and authority.
Governance example. A defined event triggers enrolment pause and review. Every submitted document now tells the same operational truth.
Seek approval for required amendments before implementing them, except for any authorised urgent safety action.
Report deviations and incidents
Frame the process for departures, breaches and adverse events as part of one coherent protocol, not an isolated form answer.
Reviewers cannot assess the real study when minor deviations disappear from the record.
State the facts and state route, timing and corrective action.
Protocol example. A consent-version error is isolated, reported and prevented from recurring. The ethical question becomes visible.
Use the institution's current form and terminology; do not self-assign a review category.
Control amendments
Follow the participant through changes to approved materials or procedures from first contact to final disposition.
Choice is weakened when researchers improve the study before approval of the change.
Explain the actual interaction and distinguish urgent safety action from ordinary amendment.
Participant example. A new interview question waits for ethics approval before use. A person can understand what will happen before agreeing.
A signature records a decision; it does not repair unclear information or hidden pressure.
Plan results communication
Evaluate how findings reach participants and the public with benefit, burden and uncertainty kept separate.
Proportionality is obscured when dissemination promises individual diagnosis.
Name foreseeable effects and state aggregate results and limits.
Risk example. Participants may request a plain-language summary after analysis. The safeguard now responds to a stated mechanism.
Minimal risk is a category within a framework, not permission to write that no risk exists.
Align every submitted document
Protect the consistency among protocol, form, script and instruments across people, systems and time.
Confidentiality becomes a slogan when session length or sample count differs across files.
Describe access, transfer, retention and run a cross-document check.
Data example. Recruitment, consent and protocol all state two 45-minute visits. The proposed control can be reviewed and tested.
Did You Know? Singapore's Ministry of Health says the Human Biomedical Research Act clarifies roles and responsibilities so participant health, welfare and safety remain paramount. This is why the consistency among protocol, form, script and instruments must be written as a bounded, testable claim.
Write a review-ready conclusion
Align the ethical case and unresolved decisions across the application, protocol, scripts and participant materials.
Approval rests on a different study when the summary claims the study is ethical.
Run the cross-document check and show value, proportionality, safeguards and questions.
Governance example. The conclusion identifies one residual privacy risk and its oversight. Every submitted document now tells the same operational truth.
Seek approval for required amendments before implementing them, except for any authorised urgent safety action.
A worked example
A student proposes interviews about academic pressure. The application says participants face no risk, the poster promises useful coping insights, the consent form says data are anonymous and the protocol records names beside audio files on a personal laptop.
The revised submission maps recruitment through debrief, separates voluntary participation from lecturer authority, recognises possible distress, provides a pause-and-support pathway, describes coded rather than anonymous data, uses approved storage, states withdrawal limits and aligns every document.
English becomes an ethical control. It lets a potential participant know what will happen and lets the reviewer see whether value, choice, risk and safeguards genuinely fit together. Clear writing does not decorate ethics; it makes accountable review possible.
A practical checklist
- Correct review route confirmed
- Responsible institution and investigator clear
- Question and design aligned
- Population and criteria justified
- Recruitment and power relationships addressed
- Information and consent understandable
- Burden, risk and benefit honest
- Privacy and confidentiality accurate
- Data and samples governed
- Conflicts and monitoring disclosed
- Incidents and amendments controlled
- All documents consistent
Advice for students, parents and young adults
Students can practise by drawing a participant journey before completing any form. Every invitation, question, recording, transfer and follow-up should appear somewhere in the application.
Parents supporting a young researcher can ask a useful question: would a participant understand what happens, what choice they have and whom to contact, without needing the researcher beside them?
This guide does not determine whether a project needs review or which law applies. Researchers must use the current process of their institution, sponsor and jurisdiction and must not begin covered recruitment or data collection before required approval.
Frequently asked questions
Is every student project human biomedical research?
No. Scope depends on the activity, institution and applicable framework. Obtain an official determination rather than self-exempting.
Can recruitment begin while approval is pending?
Not when the applicable process requires prior approval. Follow the institution's written rules and decision.
Does a signed form prove informed consent?
No. Consent is a communication and decision process; the signature documents part of that process.
Are coded data anonymous?
Usually not when a key or realistic re-identification route remains. Use the institution's approved terminology.
Should the application say there is no risk?
Avoid absolute claims. Describe foreseeable risks, mitigations and residual uncertainty proportionately.
Do changes after approval need review?
Many changes require an approved amendment before implementation, apart from any permitted urgent action needed to remove immediate hazard.
The deeper English lesson
Research-ethics English is participant-journey language. It aligns intention with observable procedure, separates possibility from promise and makes power, burden, uncertainty and accountability visible before a person is asked to take part.
Useful next reading
Continue with reading an academic integrity policy, explaining scientific evidence, writing a research data management plan, writing a laboratory notebook entry, and the How English Works.