Two organisations can both say:
We follow good practice.
That sentence sounds reassuring.
It tells us almost nothing.
Good according to whom?
Practice measured how?
What counts as acceptable?
What would count as failure?
What happens if two teams interpret “good” differently?
This is the problem standards try to solve.
A standard is English used to turn a shared expectation into a repeatable benchmark.
It does not guarantee excellence.
It creates a common reference point.
That reference point can help different people build compatible things, assess similar work in similar ways, test whether a requirement has been met, and avoid reinventing the meaning of “acceptable” every time a new case appears.
Quick Read
One-sentence answer: a standard works when a recognised community or body turns shared knowledge into stable, repeatable rules, guidelines or characteristics that different users can interpret consistently and, where requirements exist, test for conformity.
ISO defines a standard as a document established by consensus and approved by a recognised body that provides rules, guidelines or characteristics for common and repeated use, aimed at achieving an appropriate degree of order in a given context.
That definition contains several ideas worth slowing down for:
- document: the benchmark has a durable representation;
- consensus: it is not merely one person’s private preference;
- recognised body: authority is attached to a defined institution or standards process;
- common and repeated use: the same benchmark can be applied across many cases;
- rules, guidelines or characteristics: the document can define mandatory criteria, recommended practice or stable properties;
- order: the aim is to reduce incompatible interpretations and unnecessary variation.
The central English problem is therefore not only:
What should we say?
It is:
How do we write a benchmark so different readers can make the same distinction between conforming and non-conforming cases?
The Standard Is Not the Policy
A policy belongs to an organisation’s chosen governance.
A standard often reaches beyond one organisation.
Policy:
Our organisation requires all assessment reports to use the approved accessible document format.
Standard:
The document format must conform to the specified technical and accessibility requirements.
The policy tells one organisation what it chooses or requires.
The standard provides a common benchmark against which a format, process or output can be described or assessed.
policy = organisational rule
standard = shared benchmark
An organisation can adopt a standard inside its policy. The two layers cooperate without becoming the same thing.
The Standard Is Not the Specification
A specification usually defines what one target product, service, system or deliverable must fulfil.
A standard can provide the shared framework from which many specifications draw.
Standard:
Colour contrast must meet the defined accessibility criterion.
Specification:
The eduKateSG article template must use heading and body combinations that meet the required contrast threshold in every supported theme mode.
The standard establishes the benchmark.
The specification applies requirements to a particular target.
standard = common rule space
specification = target-specific requirement set
Consensus Does Not Mean Everyone Gets Everything They Want
ISO describes consensus as general agreement characterised by the absence of sustained opposition to substantial issues, developed through a process that seeks to consider the views of the parties concerned.
Consensus is therefore not identical to unanimity.
That matters because standards exist precisely where many interests need one shared reference.
Manufacturers want feasibility.
Users want reliability.
Regulators may want safety.
Accessibility advocates want inclusion.
Technical experts want precision.
A standard tries to produce language stable enough to survive those competing pressures.
Consensus is not the absence of disagreement.
It is the disciplined construction of a shared enough rule.
Requirements, Recommendations and Permissions Need Different Words
ISO’s normative language makes an important distinction.
- shall indicates a requirement;
- should indicates a recommendation;
- may indicates permission;
- can indicates possibility or capability.
These words are tiny.
Their institutional force is not.
Compare:
The system shall retain an audit record.
The system should display the audit record in chronological order.
Administrators may export the record.
The system can support up to 10,000 entries.
The first sentence establishes a conformity condition.
The second recommends a preferred practice.
The third grants permission.
The fourth describes capability.
If these modal verbs are treated as decorative synonyms, conformity becomes unstable.
Normative language is a force map.
A Requirement Needs to Be Verifiable
ISO describes a requirement as objectively verifiable criteria that must be fulfilled for conformity to be claimed.
Weak:
The interface shall be attractive.
Attractive according to whom?
How would two evaluators determine the same result?
Stronger:
All primary navigation labels shall remain visible at 200% browser zoom without requiring horizontal scrolling at the specified viewport width.
The second statement contains observable criteria.
It may still require a test method.
But the reader can at least imagine a pass or fail.
A standard becomes operational when its words can meet evidence.
Conformance Is a Claim About Relationship
When something “conforms to a standard,” the claim is not that the object is perfect.
It is that the object fulfils the specified requirements relevant to that conformity claim.
This creates a relationship:
standard requirements → target object → evidence → conformity judgement
ISO notes that conformity assessment can be carried out by first parties, second parties or independent third parties.
That distinction matters because:
- a manufacturer can test its own product;
- a purchaser can test a supplier’s product;
- an independent body can assess conformity separately.
The same written requirement may therefore be read by people with different incentives.
Ambiguous language becomes especially expensive when one side benefits from a loose interpretation.
Definitions Stabilise the Vocabulary of the Standard
Standards often begin by defining important terms.
Why?
Because ordinary words may not be precise enough for conformity.
“User.”
“System.”
“Accessible.”
“Failure.”
“Available.”
Each may need a controlled local meaning.
This is the same reason Policy needs definitions and Glossary needs local agreements.
If two readers use the same word for different states, the standard may look shared while the test remains private.
Numbers Need Units, Conditions and Tolerances
Standard:
Response time shall be less than 2 seconds.
Under what load?
Measured from which event to which event?
On which hardware?
For every request or 95% of requests?
Numbers create false precision when measurement conditions remain undefined.
More robust standard language might specify:
- unit;
- measurement method;
- environment;
- tolerance;
- sample size;
- acceptable proportion;
- rounding rule.
The number is only one part of the criterion.
Standards Reduce Unnecessary Variation, Not All Variation
Standardisation is often misunderstood as making everything identical.
That would be disastrous in many domains.
A standard can define the interface while allowing many implementations.
It can define minimum safety while allowing many designs.
It can define file structure while allowing many kinds of content.
A good standard fixes the dimensions that need agreement and leaves freedom where variation is useful.
This is one reason standards can support innovation rather than only constrain it.
If everyone agrees on the socket, designers can innovate on what plugs into it.
Interoperability Is One of the Great Powers of Standards
Imagine every railway using a different track width.
Every charger using a different connector.
Every file using a private structure.
Every laboratory recording units differently.
People can still build things.
Those things struggle to work together.
Standards often create agreed interfaces between otherwise independent systems.
shared representation → compatible interpretation → lower coordination cost
This applies to English too.
When different organisations use the same defined fields, units, labels or document structures, information crosses boundaries with less translation.
The standard becomes a shared grammar between systems that do not share an owner.
References Can Make One Standard Depend on Another
A standard does not need to repeat every rule it depends on.
It may refer to another standard.
This creates a dependency graph.
Standard A requires conformance with Standard B for one component.
That saves duplication.
It also creates version questions.
Does Standard A refer to:
- a specific edition of Standard B;
- the latest edition;
- a particular clause only?
Reference language should make that relationship clear enough that two readers do not inherit different dependency sets.
Revision Is Part of Standardisation
A standard is stable.
It is not frozen forever.
Technology changes.
Evidence improves.
New risks appear.
Old tolerances become inappropriate.
Revision creates another language problem:
Which version of the shared benchmark are we talking about?
Version, publication date and supersession therefore matter.
“Conforms to Standard X” can be incomplete if several editions exist and the differences matter.
The timestamp and version number become part of the claim.
A Standard Can Become Too Weak
Suppose a standard says:
Systems should be secure.
The sentence sounds sensible.
It may be almost impossible to use consistently.
Weak standards suffer from:
- undefined terms;
- unmeasurable adjectives;
- missing conditions;
- unclear normative force;
- conflicting clauses;
- requirements with no practical evidence path.
A standard can therefore exist formally while producing little order in practice.
The authority of a standard cannot compensate for language that refuses to distinguish pass from fail.
A Standard Can Also Become Too Rigid
The opposite danger is overspecification.
A standard can freeze one implementation when the real need was only to standardise the interface or outcome.
For example:
Every compliant system shall use Button X in Location Y.
Perhaps the real requirement was:
Users shall be able to reach the primary action through a clearly labelled control accessible by keyboard and pointer input.
The second statement standardises the outcome rather than one visual implementation.
Good standards choose the correct level of abstraction.
Voluntary Does Not Mean Irrelevant
ISO notes that its International Standards are voluntary and do not replace national law.
Yet voluntary standards can still become powerful.
A contract may require them.
A purchaser may prefer conforming suppliers.
A regulator may refer to a standard.
An industry may treat one as the normal interoperability baseline.
The English distinction matters:
voluntary standard ≠ legally mandatory rule
voluntary standard can still become operationally important
Readers should ask what gives the standard force in the situation they are examining.
Primary School: Make a Fair Shared Standard
Young learners can practise standard thinking through a classroom poster task.
Weak benchmark:
Your poster should look good.
Ask students to turn that into shared criteria:
- title can be read from two metres away;
- all required facts are present;
- sources are named;
- text does not cover the images;
- spelling is checked.
Now different students can understand the same benchmark.
The lesson is not bureaucracy.
It is fairness through explicit criteria.
Lower Secondary: Rewrite Vague Requirements
Give students:
- The report should be clear.
- The presentation must not be too long.
- Images should be good quality.
- The argument should be strong.
Ask them to make each statement more verifiable.
They will discover that some concepts cannot be reduced to one number without losing meaning.
That tension is valuable.
Standards require precision, but precision must still represent what actually matters.
Upper Secondary: Audit Normative Language
- Who established the standard?
- What gives it authority in this context?
- Which clauses are requirements?
- Which are recommendations?
- Can the requirements be verified?
- Are key terms defined?
- Are measurement conditions clear?
- Does the standard overcontrol implementation?
- Which version applies?
- What evidence supports a conformity claim?
This turns standard reading into a lesson in modal verbs, measurement, institutional authority and evidence.
Ten Failure Modes of Standard English
- Consensus theatre. A private preference is presented as though it were a shared benchmark.
- Normative-force drift. Shall, should, may and can are used without consistent meanings.
- Unverifiable requirement. “High quality” or “easy to use” appears without testable criteria.
- Definition drift. Key terms carry different meanings across readers.
- Measurement ambiguity. Numbers appear without units, conditions, tolerances or methods.
- Conformance inflation. Meeting selected requirements is presented as proof of overall excellence.
- Version ambiguity. A conformity claim does not identify which edition applies.
- Dependency ambiguity. Referenced standards or clauses are unclear.
- Overspecification. The standard freezes implementation details when only the outcome needed agreement.
- Under-specification. The document sounds authoritative but leaves too much interpretation for consistent application.
How to Write a Better Standard
Start with the recurring coordination problem.
- What needs common agreement?
- Which dimensions must be standardised?
- Which variation should remain free?
- What terms need definitions?
- Which clauses are requirements, recommendations or permissions?
- Can every requirement meet observable evidence?
- What measurement conditions or tolerances are needed?
- How will conformity be assessed?
- Which external standards are dependencies?
- How will revision and versioning be handled?
Then perform the independent-reader test:
Could two competent parties with different incentives read this standard, test the same object and reach the same conformity judgement for the same reasons?
The Deeper Idea: Standards Are Shared Memory About What Counts
Without a standard, every new interaction begins with negotiation.
What format?
What unit?
What minimum quality?
What does failure mean?
A standard stores some of those decisions in advance.
past negotiation → shared standard → future compatibility
This is why standards can make civilisation feel invisible when they work well.
Doors fit frames.
Files open.
Units agree.
Tests mean something similar in different places.
English helped turn those agreements into durable shared rules.
Reader Checklist
- Who created this standard?
- What recurring problem does it solve?
- Which clauses are mandatory?
- Are requirements verifiable?
- Are terms and units defined?
- What does conformity actually prove?
- Which edition applies?
- Does the standard allow useful variation?
- What other standards does it depend on?
- Could different evaluators reach the same judgement?
Related eduKateSG Reading
- How English Works | The Policy
- How English Works | The Procedure
- How English Works | The Specification
- How English Works | The Glossary
- How English Works | The Checklist
Research and Further Reading
- ISO — Useful Definitions for Standards and Standardisation
- ISO — How Normative Language Works in Standards
- ISO 19105 — Conformance and Testing Terminology
Final idea: a standard is good when it turns a recurring argument about what counts into a benchmark clear enough that strangers can coordinate without first becoming the same organisation.