An evidential marker is a grammatical form, word or expression that indicates something about the source of information behind a statement. Compare ‘I saw the door open’, ‘A witness reported that the door opened’ and ‘The door apparently opened’. These sentences are not interchangeable: the first describes direct visual experience, the second attributes a report, and the third may suggest an inference or reported information whose precise interpretation depends on context.
Vocabulary becomes a powerful verification tool when students can distinguish what a writer directly observed, what someone else said, what is inferred from clues and what remains unknown. Words such as reportedly, according to, apparently, seems and I observed help readers locate claims in the chain from evidence to conclusion. They do not magically make the underlying source reliable; they show how the speaker presents access to the information.
This guide explains evidentiality as a linguistic concept, how languages differ in marking it, and how English uses varied lexical and grammatical strategies. It then develops original examples for Science reports, English comprehension, Mathematics explanations, news literacy, school messages and digital summaries. You will learn when to use an attribution, how to separate evidence source from confidence and how to write without inventing observations.
Source note: Every pupil, report, observation and numerical result used in the examples is fictional. This is a guide to interpreting information-source expressions, not a claim about actual institutions or events.
Contents: follow evidence from source to claim
Explore all 42 chapters and exercises
- One event, three information routes
- The technical definition and source literature
- Evidentiality is not the same as certainty
- Direct visual evidence
- Other sensory evidence
- Reported information
- According to and its source
- Reportedly and source uncertainty
- Allegedly and contested claims
- Purportedly and supposedly
- Apparently can have more than one reading
- Evidently may express an inference
- Seemingly and appearance
- I infer versus I observe
- A statement can have several evidence sources
- Quotation is a source relationship, not endorsement
- Hearsay is not always worthless
- The link to epistemic modality
- An information source does not establish causation
- The primary source may differ from the writer
- Source chains can break
- Source date can change whether evidence remains usable
- A screenshot proves only certain things
- A citation is not an argument
- Source and confidence in a two-column grid
- Worked Science case: the water in the cup
- Worked Mathematics case: reading an answer key
- Worked English case: motive from a story
- Worked group case: the homework deadline
- An AI-generated summary needs source checking
- Misuse of ‘obviously’ and ‘clearly’
- Evidential markers can be strategically misleading
- Evidential markers and discourse markers do different jobs
- Evidentiality in other languages
- Translation can lose an information-source distinction
- A source route can be more important than a fancy adjective
- The practical source-check ladder
- A small classroom workshop
- Diagnostic: five common errors
- Ten questions and answer explanations
- Frequently asked questions
- Sources, limitations and next action
1. One event, three information routes
Consider a fictional classroom incident: a window is open. Mira saw it open, Kai was told it had opened, and Alicia inferred that it had opened because she saw a moving curtain.
All three may discuss a similar event, but their access to information is different. Direct observation, report from someone else and inference from clues should not be silently merged.
Ask which person could truthfully write ‘I witnessed the window being opened’, which could write ‘Kai told me the window was open’ and which needs an appropriately bounded inferential statement.
2. The technical definition and source literature
Alexandra Aikhenvald’s Evidentiality (Oxford University Press, 2004) explains grammatical marking of information source, including witnessed, reported and inferred information. Some languages encode such distinctions through obligatory grammatical categories.
English does not generally use one obligatory grammatical evidential paradigm for every assertion. Instead, speakers often express source information through verbs, adverbs, modal constructions and attribution phrases. Researchers debate the degree of grammaticalisation of some English forms.
A classroom should therefore distinguish evidentiality in its stricter grammatical sense from practical English evidential expressions. The shared reasoning question is: how does this claim indicate where its information comes from?
3. Evidentiality is not the same as certainty
‘I saw the flash’ identifies sensory access, but does not automatically specify how certain the witness is about its cause. ‘I am sure the flash was lightning’ expresses confidence in an interpretation.
An expression can signal a source, degree of certainty or a mixture, depending on its meaning and use. We should not force every marker into only one category.
Ask two separate questions: ‘What information route is claimed?’ and ‘How strongly does the speaker commit to the proposition?’ The answers can differ.
4. Direct visual evidence
‘I watched the paper fall from the table’ claims visual observation of an event. If the speaker did not see it fall but found it on the floor afterward, that sentence would misrepresent the evidence route.
A more accurate report might say ‘I found the paper on the floor; I did not see it fall.’ The wording protects the difference between observation and later inference.
Teach learners to avoid reporting what they imagined happened as something they actually witnessed.
5. Other sensory evidence
‘I heard a knock’ describes auditory experience. It does not by itself establish who produced the sound or why.
Smell, touch and other sensory information can similarly support an observation while leaving its cause uncertain. A student may notice an unfamiliar smell without knowing the material responsible.
An honest account separates the sensory signal from any explanation it suggests. Good writing need not turn every sensation into a confident causal claim.
6. Reported information
‘The class representative said the session begins at three’ attributes a timetable claim to that speaker. It does not independently verify the schedule.
Attribution identifies who supplied the information. The reader may still need the current official timetable if a practical decision depends on the detail.
Ask pupils to distinguish the fact of a report being made from the truth of what the report asserts. Both can be discussed without automatically endorsing the content.
7. According to and its source
‘According to the school notice, the workshop is on Friday’ names a documentary source. The credibility of the statement depends on whether the notice is genuine, current and being interpreted correctly.
An unattributed ‘according to reports’ leaves the source much less precise. Sometimes confidentiality or genre limits attribution, but the limitation should be clear where consequential claims are made.
The phrase supports traceability; it does not by itself certify the statement as correct.
8. Reportedly and source uncertainty
‘The equipment was reportedly delivered’ presents delivery as information from reports, not necessarily direct observation by the writer.
The word may appropriately limit commitment while identifying an indirect information route. Yet an unspecified report can still be inaccurate.
Ask ‘Reported by whom, on what date and with what confirmation?’ A practical follow-up may require checking a delivery record rather than repeating reportedly several times.
9. Allegedly and contested claims
‘A person allegedly broke a rule’ represents an allegation without establishing that the wrongdoing occurred. It must not be silently paraphrased as ‘The person broke the rule’.
The same caution applies when students read a news account or fictional dispute. Allegations, investigations, findings and final decisions are distinct.
Use precise language without making unsupported accusations. A marker can signal an unresolved claim but is not a substitute for fair evidence assessment.
10. Purportedly and supposedly
These expressions often signal that a description or claim is presented without full endorsement by the speaker. Their connotations differ by context and may carry scepticism.
Do not replace apparently, purportedly and reportedly mechanically. Test a candidate against the real source and intended degree of endorsement.
Consult usage examples when the nuance matters. A writer should not use a sceptical-sounding marker merely to make a sentence dramatic.
11. Apparently can have more than one reading
‘Apparently the bus has been delayed’ might be based on information received or on evidence the speaker interprets. The actual route depends on context.
Research on English evidential adverbs analyses these variations. The Cambridge English Language & Linguistics study on evidential adverbs distinguishes reportative and inferential uses across constructions.
Where a practical decision matters, an explicit source statement may be clearer: ‘The bus operator’s service notice reports a delay’ or ‘I can see a delay indicator on the display’.
12. Evidently may express an inference
‘Evidently the answer was copied incorrectly’ may imply an inference from the work shown, not a directly observed copying event.
Before choosing the adverb, identify which clue supports the inference. If both the input and written output are visible, the writer can describe their difference without inventing when the error occurred.
An apparently persuasive evidential word does not prove the conclusion. Explain the evidence and acknowledge relevant alternatives.
13. Seemingly and appearance
‘The two diagrams seem identical’ describes an impression that may be revised after closer comparison. ‘The diagrams are identical’ makes a stronger assertion.
The word can help a student communicate uncertainty appropriately without abandoning careful checking.
Add what remains to be tested: dimensions, labels or underlying data. A qualified claim is useful when the next verification step is identifiable.
14. I infer versus I observe
An observation records something accessible through the relevant method. An inference uses observations or premises to reach a further conclusion.
‘The ruler reading was 12 cm’ may describe a measurement. ‘The object probably expanded because it warmed’ proposes an explanation dependent on additional evidence.
A good Science report identifies the recorded data and the reasoning route separately. Neither a long technical word nor a neat graph replaces the required justification.
15. A statement can have several evidence sources
A teacher checks an earlier note, the current worksheet and a pupil’s explanation before giving feedback. One conclusion may rest on several sources.
A single evidential adverb may not convey all this. A short sentence such as ‘I compared the two versions and asked the learner to explain the changed line’ can be more transparent.
Use source labels proportional to the task. Some claims require multiple checks; others need only direct reference to an uncomplicated record.
16. Quotation is a source relationship, not endorsement
A report quotes a pupil saying, ‘The exercise was impossible.’ The reporter can truthfully state that the pupil said this, while the evaluation of the exercise remains open.
Quotation marks identify attributed wording under an accurate reporting practice. They do not prove that the quoted judgement is universally correct.
Teach students to separate someone said P from P is true. This is a practical entailment distinction.
17. Hearsay is not always worthless
Secondhand information can be valuable when it comes through reliable, verifiable records or experts. Direct observation can also be mistaken.
The point of evidential marking is not to rank every observation above every report. It is to reveal the source route so that appropriate checking becomes possible.
Ask whether the report names a credible source, whether the original information can be inspected and what the consequences of error are.
18. The link to epistemic modality
Expressions such as might, must and probably can indicate possibility, necessity or degree of commitment, depending on context. Evidentiality centres on the source or type of evidence.
These functions can overlap. Researchers do not agree that every form belongs cleanly to one category in every construction. Aikhenvald’s work argues for distinguishing grammatical evidentiality from epistemic modality.
For accessible English teaching, show one sentence specifying a source and another specifying uncertainty, then allow a third that combines both.
19. An information source does not establish causation
‘The report states that scores increased’ attributes a claim. It does not establish which action caused the change.
‘I observed greater growth in one pot’ records a comparison, not a controlled explanation. A causal assertion would require additional methodological and evidential support.
Use the Entailment Test guide to inspect whether a conclusion extends beyond the recorded source.
20. The primary source may differ from the writer
An essay says, ‘A website reports that the museum opened in a particular year.’ The website may itself be summarising a museum record or another publication.
If the date matters, the student can trace the claim closer to an authoritative original document where appropriate.
Create a chain: current writer → cited page → original evidence. An evidential phrase at one step does not remove the need to check subsequent steps.
21. Source chains can break
A copied quotation appears in several articles without a link to the original context. Repetition across websites may appear like independent confirmation when all copies trace back to one unchecked statement.
A careful report identifies this limitation rather than counting the copies as separate evidence.
Vocabulary such as original source, quotation, attribution and independent corroboration helps students ask sharper research questions.
22. Source date can change whether evidence remains usable
A school timetable from last year may be an authentic document yet not the current timetable. Evidence of a past schedule is not sufficient evidence for today’s schedule.
Record the date and version relevant to the question. A current official update may supersede a previously correct notice.
An evidential statement must connect the source not only to the claim but also to the period and scope where the claim is intended to hold.
23. A screenshot proves only certain things
A screenshot can show what appeared on a screen at the time it was captured. It may not establish that a backend action succeeded or that the screen represented the current state of another system.
A learner who sees ‘Saving…’ should not record ‘Saved’ solely on that basis. The visible message and confirmed outcome are different facts.
Name exactly what the evidence shows. Avoid upgrading an interface cue into a stronger claim without a suitable receipt or record.
24. A citation is not an argument
A sentence may include an impressive reference but misinterpret the source or extend beyond its scope. The existence of a citation does not establish that the conclusion follows.
A pupil can check the source, extract the relevant claim, note uncertainty and explain how it supports the point being made.
This practice links evidential markers to reasoning rather than decorative bibliographies.
25. Source and confidence in a two-column grid
Create two headings: ‘How the information was obtained’ and ‘How strongly the claim is supported’. A direct measurement, a report and an inference belong in the first column; the second records methodological and interpretive limitations.
For an invented graph result, the first column might say ‘read the labelled value from the provided chart’; the second might say ‘supports this plotted value, but not the proposed cause’.
The columns separate information route from confidence. They are an educational aid, not a validated measure of evidence quality.
26. Worked Science case: the water in the cup
A fictional child finds less water in a cup at the end of a lesson. They did not watch the entire interval. They write, ‘I saw the water evaporate.’
The statement overstates direct observation. A more accurate record is ‘The measured water level was lower at the later check. Evaporation is one possible explanation under the relevant conditions.’
Discuss alternative causes, measurement uncertainty and what further evidence would distinguish them. A precise evidential phrase makes an incomplete investigation more honest, not less scientific.
27. Worked Mathematics case: reading an answer key
A student writes, ‘The answer key says x = 5.’ That sentence accurately attributes the key’s claim but does not establish the answer’s correctness if the key itself contains an error.
Substitution into the original equation provides a different kind of support. If the proposed x satisfies the given equation under its conditions, the learner can report that check.
Keep ‘reported by the key’ separate from ‘verified by calculation’. A vocabulary distinction directs the next useful action.
28. Worked English case: motive from a story
A fictional character leaves a room suddenly. The pupil writes, ‘The character was frightened because the passage says she left.’
Leaving may be consistent with fear, but the action alone does not necessarily establish that motive. The pupil can write ‘The character may have been frightened’ if other cues support the possibility, while acknowledging alternatives.
The evidence source is the passage; the proposed psychological explanation is an inference. The distinction protects comprehension answers from unsupported certainty.
29. Worked group case: the homework deadline
One student says the assignment is due Thursday because an old screenshot shows Thursday. Another says Friday because they remember a classroom announcement.
Neither recollection alone resolves the current official deadline. The group needs the relevant current instruction or an authorised clarification.
The correct status might be ‘Two conflicting reports; deadline unverified’. That phrasing is more actionable than choosing the more confident speaker.
30. An AI-generated summary needs source checking
A machine-generated summary can present a claim confidently without showing the underlying evidence. The wording should not be treated as an eyewitness account or an authoritative citation.
Open and inspect relevant original sources where the claim matters; distinguish the summary’s paraphrase from source wording and verify the date, context and quantitative details.
The action principle remains the same across human and digital writing: source attribution is a starting point for evaluation, not proof of accuracy.
31. Misuse of ‘obviously’ and ‘clearly’
‘Clearly, the new method works’ may express high confidence without identifying the evidence needed for that conclusion.
Replace it with a bounded statement naming the observed result and the scope of the comparison. If the data support only a local difference, report that difference rather than claiming universal effectiveness.
Such words can have legitimate rhetorical and inferential functions; the problem arises when they substitute for explanation.
32. Evidential markers can be strategically misleading
A writer may repeatedly say ‘experts say’ without naming the experts, their relevant expertise or what they actually concluded.
The phrase may create an impression of authority while leaving the source impossible to evaluate. Another writer might use ‘apparently’ to avoid responsibility for an unsupported insinuation.
Ask for a traceable source, accurate quotation and meaningful limitation. A signal of evidence should be evaluated as a claim about evidence.
33. Evidential markers and discourse markers do different jobs
‘However‘ often signals a contrast between discourse segments; ‘reportedly‘ often attributes information to reports. Both influence interpretation but do not normally announce the same relationship.
A writer could use both: ‘The club reportedly changed the meeting time. However, the current official schedule still lists the original session.’
The Discourse Marker guide explains how linking expressions guide movement between ideas.
34. Evidentiality in other languages
Some languages grammaticalise evidential distinctions, so a statement may require marking whether information is witnessed, reported or inferred. The number and exact categories differ across languages.
An English learner should not assume all languages work like English or that every foreign-language marker maps to one familiar English adverb.
The Aikhenvald monograph gives the broader typological background. Accurate translation requires attention to the particular language’s grammatical system.
35. Translation can lose an information-source distinction
If a source sentence grammatically marks hearsay, an English translation that states the underlying event as an unqualified fact may remove an important distinction.
An English phrase such as ‘reportedly’ or an explicit attribution might preserve it, depending on context and style. The correct choice needs analysis of the actual source form and use.
Do not apply a word-for-word mapping mechanically. Preserve the source’s evidential relationship with natural target wording where possible.
36. A source route can be more important than a fancy adjective
A pupil is tempted to replace ‘I read’ with ‘It is incontrovertible’. The latter does not disclose the information source and vastly strengthens the claim.
A precise explanation—’The published report states this figure, but I have not checked the underlying dataset’—can be longer and more trustworthy.
Vocabulary competence is not simply using more advanced words. It is matching the claim’s strength to its evidence and communicating its limits.
37. The practical source-check ladder
Begin with what was claimed. Next identify who or what supplies the information, what was directly observed, what was inferred and which parts remain uncertain.
Where relevant, check the original wording, date, audience and scope. Finally, choose an expression whose evidential commitment matches the available support.
This ladder is an adaptable teaching routine. It should not be mistaken for an official scientific protocol or sufficient verification for high-stakes conclusions.
38. A small classroom workshop
Provide three fictional descriptions of one event: an eyewitness observation, a secondhand report and an inference from an after-effect. Do not tell pupils which is the most trustworthy automatically.
Ask them to label information routes, describe what each statement actually establishes and choose one further check that could settle a disagreement. Then rewrite one overconfident claim with accurate attribution.
Look for independent transfer to a new paragraph. A pupil who can classify a rehearsed example may still need practice using the method without the labels.
39. Diagnostic: five common errors
Error one: treating reportedly as confirmation. Error two: calling an inference a direct observation. Error three: equating high speaker confidence with independent verification. Error four: citing a source whose wording does not support the claim. Error five: using an outdated authentic document as proof of a current fact.
Each error needs a specific repair: identify the actual reporting source, revise the information route, inspect evidence, correct the interpretation or find the current record.
The vocabulary should tell the learner what to investigate, not merely that the sentence sounds uncertain.
40. Ten questions and answer explanations
1. Does ‘reportedly’ guarantee that the report is true? No. 2. Does ‘I saw’ establish the cause of every observed event? No. 3. Is information source identical to certainty? No. 4. Can English mark evidential meanings without an obligatory grammatical suffix? Yes.
5. Can a citation be inaccurate? Yes. 6. Is an inference the same as direct witness? No. 7. Does ‘allegedly’ establish guilt? No. 8. Can several copied sources derive from one report? Yes.
9. What should happen after conflicting reports? Identify and check the relevant sources. 10. What makes an evidential phrase useful? It communicates the information route and allows its claims to be assessed at an appropriate level.
41. Frequently asked questions
Is evidentiality only a language topic? It is a linguistic category, but recognising information routes helps with ordinary reading, research and reporting. Does English have evidential markers? English has a variety of expressions indicating sources and inferences, though not generally an obligatory grammatical paradigm covering every assertion.
Are apparently and reportedly synonyms? Their uses can overlap but differ; context matters. Should students always sound uncertain? No. They should state supported observations clearly and qualify conclusions when evidence warrants it.
Why is this vocabulary technology? Source-marking words help readers distinguish observation, quotation, report and inference before deciding what to trust or check.
42. Sources, limitations and next action
Alexandra Y. Aikhenvald, Evidentiality (2004), provides the grammatical and typological foundation. See Eric Mélac’s research on evidentiality in English and the study of English evidential adverbs for distinctions between lexical source markers and inferential constructions.
All classroom observations in this article are invented teaching examples. The linguistic research does not validate specific pupil outcomes or endorse one fixed source-check worksheet for every discipline.
Choose a sentence from a homework explanation or public source. Write how the speaker knows it, whether the claim is directly observed, reported or inferred, and which part still needs checking. Then select wording that communicates this status accurately. You have used vocabulary to protect evidence rather than decorate the sentence.