Did You See It, Infer It, or Hear About It? — What Matsés Reveals About the Grammar of Evidence

An assertion tells us what someone claims. Its source tells us what kind of claim we are dealing with.

— Tymur Levitin

Imagine arriving at a house and discovering that the door is open.

You say:

Someone entered the house.

Your companion asks:

Did you see anyone enter?

No.

Did someone tell you?

No.

Then how do you know?

You point to the open door, muddy footprints and a broken lock.

You did not witness the event.

You reconstructed it from evidence.

Now imagine a different situation.

You saw someone enter.

Or a neighbour told you someone entered.

Or you simply suspect that someone entered because the owner has been away.

The statement may look identical:

Someone entered the house.

But the grounds for making it are different.

English allows us to express those differences:

I saw someone enter.

Someone apparently entered.

I was told someone entered.

I suspect someone entered.

Yet an ordinary English statement does not always require us to specify which situation applies.

We can simply say:

Someone entered the house.

And leave the listener to work out how we know.

In Matsés, a language spoken in Amazonian Peru and Brazil, the grammar of past-event reporting works differently.

The speaker is normally required to distinguish different sources of information.

And in some constructions, the language goes further.

It can encode not only when an event happened, but also when the evidence for that event was encountered.

This makes Matsés an extraordinary case study in the relationship between grammar, evidence and responsibility for what we say.

But it also presents a danger.

We must not confuse grammatical evidence marking with proof.

A language can require speakers to identify their information source without guaranteeing that every statement is true.

That distinction will guide our investigation.

1. Four statements that are not equivalent

Consider these English sentences:

1. I saw him leave.
2. He must have left.
3. Someone told me he left.
4. I think he probably left.

All four concern the same possible event.

A person leaving.

But they differ in how the speaker acquired or evaluated the information.

In the first case, the speaker claims direct observation.

In the second, the speaker presents a conclusion.

In the third, the information comes from another person.

In the fourth, the speaker expresses a judgment with uncertainty.

These distinctions matter in everyday life.

They matter in journalism.

They matter in medicine.

They matter in legal proceedings.

They matter in education.

And they matter whenever someone must decide whether to act on information.

But English does not always make them grammatically obligatory.

Matsés offers a different arrangement.

2. What is evidentiality?

In linguistics, evidentiality concerns the grammatical expression of information source.

The relevant question is:

How did the speaker obtain the information being communicated?

Possible sources include direct sensory experience, inference from observable evidence and reports from other people.

Some languages distinguish several such sources through grammatical markers.

Others rely more heavily on lexical expressions, context or optional constructions.

English can say:

I saw it.

Apparently.

According to Maria.

They say that…

These expressions are meaningful.

But English generally does not require a specialized evidential ending on every past-tense verb.

Matsés has a much more elaborate grammatical system for reporting past events.

Its morphology makes source-of-information distinctions central to the construction of those statements.

3. The Matsés language

Matsés belongs to the Panoan language family.

It is spoken by Matsés communities in the Amazonian region of Peru and Brazil.

The language has been studied extensively by linguist David W. Fleck.

His research on Matsés evidentiality is especially important because it describes an unusually elaborate system combining information source with temporal distinctions.

The central study is:

David W. Fleck (2007), Evidentiality and Double Tense in Matses.

Fleck demonstrates that Matsés speakers reporting past events must normally select grammatical constructions that distinguish how the information was obtained.

The relevant system includes:

• Experiential: direct experience of the event or state.
• Inferential: inference from evidence resulting from an event.
• Conjecture: an assumption or speculation without the relevant concrete evidence.
• Hearsay: information conveyed through a quotation-based reporting construction.

An important grammatical qualification is necessary.

The first three involve specialized evidential inflection.

Hearsay is expressed through a quotative construction rather than an equivalent reportative suffix.

Functionally, however, it belongs to the system of choices speakers use when reporting past events.

That distinction matters because a language’s communicative categories and its morphological categories do not always have identical boundaries.

4. Seeing an event and seeing its consequences

Imagine two hunters.

The first sees an animal cross a path.

The second arrives later and sees its tracks.

Both may conclude:

An animal passed here.

But their experiences are different.

The first witnessed the event.

The second witnessed evidence left by the event.

This is a fundamental distinction in Matsés evidentiality.

Direct experience and inference are not interchangeable.

A speaker who observed the event has a different grammatical basis for reporting it from someone who reconstructed it through traces.

Notice that the second speaker may be completely correct.

The tracks may provide excellent evidence.

The issue is not simply whether the conclusion is true.

The issue is how the speaker came to know or believe it.

5. Experiential evidence is broader than vision

It would be inaccurate to define the Matsés experiential category simply as:

I saw it.

Direct experience can involve more than sight.

A person may hear an event.

Smell something.

Or otherwise directly experience a relevant state or occurrence.

The experiential category concerns direct access to the event or state, rather than vision alone.

This matters because everyday explanations of evidentiality often reduce it to the contrast between eyewitnesses and people who were not present.

But direct sensory experience has several possible forms.

The grammatical category must be understood according to the language’s actual rules, not an English stereotype about eyewitness testimony.

6. A documented Matsés example: recent experiential past

Fleck identifies the suffix:

-o

as a marker of recent past experiential information.

Its use indicates that the speaker has direct experiential grounds for the past event or state.

For comparison, the Matsés form:

daëd ik-o-sh

is discussed in the linguistic literature with the meaning:

There are two.

Literally, its grammatical construction uses a past experiential form.

The example is associated with answering a question about the number of wives a person has.

The interpretation can be understood approximately as:

There were two, as last directly known.

The important point is not that Matsés speakers cannot talk about the present.

It is that the relevant grammatical form encodes a relationship between a known state and the speaker’s direct experience of it.

English often presents the corresponding information without making that temporal and evidential relationship explicit.

7. Why the past matters

Why would a language develop especially elaborate evidential distinctions for past events?

One reason is that past events are often no longer directly observable.

The event has ended.

The participants may have left.

Only traces, memories or reports remain.

A speaker must therefore distinguish between different relationships to the event.

Did they experience it?

Did they encounter its consequences?

Did they hear about it?

Are they making an assumption?

These are not merely philosophical questions.

They arise constantly in ordinary communication.

Matsés grammaticalizes distinctions that are especially relevant when the event itself is no longer available for direct inspection.

This does not mean every Matsés statement about the future has the same obligatory evidential marking.

The elaborate system described by Fleck is particularly associated with reporting past events.

8. Inference is not the same as guessing

Suppose you return home and find water on the floor.

You notice that a pipe is leaking.

You conclude:

The pipe burst while I was away.

You did not witness the rupture.

But you have physical evidence.

Now imagine that you are travelling and have heard nothing about your house.

You suddenly say:

Perhaps a pipe burst at home.

That is a different epistemic situation.

The first statement is an inference from observed consequences.

The second is speculation without comparable concrete evidence.

Matsés distinguishes such categories through its grammatical system.

The difference is not necessarily the speaker’s emotional confidence.

Someone may feel extremely confident about a guess.

Another person may be cautious about a conclusion supported by evidence.

Source of information and degree of certainty are related but distinct dimensions.

9. Evidentiality is not the same as epistemic modality

This distinction deserves careful treatment.

Evidentiality concerns the source of information.

Epistemic modality concerns the speaker’s evaluation of a proposition’s possibility, probability or necessity.

Compare:

I saw the train leave.

This presents direct perceptual access.

Now compare:

The train must have left.

This presents a strong conclusion.

But the sentence alone does not tell us exactly what evidence produced that conclusion.

Perhaps the platform is empty.

Perhaps the timetable indicates departure.

Perhaps someone checked the station’s records.

Now consider:

I saw something, but I am not completely sure what happened.

Direct observation does not automatically produce certainty.

And:

I did not see the event, but the evidence is overwhelming.

Indirect evidence does not automatically produce uncertainty.

Therefore, it is incorrect to equate:

direct = certain

and:

indirect = uncertain.

Matsés makes this distinction particularly important.

10. A source marker is not a truth certificate

Imagine a speaker uses a grammatical form associated with direct experience.

Does that prove the statement is true?

No.

The speaker may have misperceived the event.

They may have misunderstood what they experienced.

They may be remembering incorrectly.

They may even be lying.

The grammar identifies the source of information being presented.

It does not independently verify the source.

This is a crucial distinction.

A witness can be sincere and mistaken.

A report can be accurate despite being secondhand.

An inference can be correct.

A confident statement can be false.

The grammatical marking of evidence is not the same thing as evidence being sufficient.

11. Three kinds of responsibility

We can separate three questions.

First: What are you claiming?

This concerns the proposition.

Second: What is your basis for claiming it?

This concerns the information source.

Third: Does that basis justify your conclusion?

This concerns the quality of the reasoning.

These questions are connected.

But they are not interchangeable.

Matsés is especially interesting because the second question is deeply integrated into the grammar of past-event reporting.

However, grammar cannot answer the third question on its own.

A language may require the speaker to identify the source category.

The listener must still evaluate whether the information is reliable.

12. The unusual problem of two times

Now we reach the most distinctive part of the Matsés system.

Imagine this sequence:

Monday: Someone builds a shelter.

Friday: You discover the shelter and examine the evidence.

Sunday: You tell another person that someone built it.

There are three relevant moments.

The event.

The discovery.

The report.

Many familiar tense systems primarily organize the event relative to the time of speaking or another reference point.

Matsés can encode a more intricate relationship.

In certain inferential constructions, the grammar distinguishes:

How long before the evidence was discovered did the event happen?

And:

How long before the present report was that evidence discovered?

This is the phenomenon Fleck calls double tense.

13. Why double tense is not simply two past events

English can say:

Someone built the shelter years ago, but I discovered it yesterday.

This clearly expresses two times.

But it does so through separate clauses and lexical time expressions.

The Matsés system can express corresponding distinctions through the morphological structure of the reported event.

The point is not that English speakers cannot understand two temporal relationships.

They obviously can.

The difference is how the grammatical system organizes them.

In Matsés, the timing of an inferred event and the timing of obtaining evidence can both become part of the verbal inflectional structure.

That is a remarkable interaction between tense and evidentiality.

14. A documented example: the shelter

Fleck provides an example involving a shelter built by non-Matsés people.

The relevant Matsés sentence is:

mayu-n bëste-wa-nëdak-o-sh

Its English meaning is approximately:

Non-Matsés people had made a shelter.

The grammatical analysis includes:

-nëdak — distant past inferential.

-o — recent past experiential.

The combination is crucial.

The speaker inferred that the shelter had been built in the distant past.

But the evidence supporting that inference was encountered recently.

The wordform therefore distinguishes the temporal location of the inferred event from the temporal location of the evidence encounter.

The speaker is not simply saying:

This happened long ago.

The grammatical construction also represents the more recent encounter with the grounds for the conclusion.

15. The discovery has its own time

Why is this so interesting?

Because the evidence itself becomes part of the temporal organization of the statement.

Imagine an archaeologist discovering an old settlement.

The settlement was constructed centuries ago.

The archaeologist discovered it last week.

The report is delivered today.

The event and the evidence encounter are temporally distinct.

English can express that difference explicitly.

But it does not ordinarily require a special combination of tense-evidential suffixes on the verb describing the original construction event.

Matsés demonstrates that a language can build this distinction into its verbal morphology.

The grammatical system does not merely locate events in time.

It can also locate the speaker’s access to evidence in time.

16. Two clocks, two questions

We can represent the relevant structure conceptually:

Clock 1: Event → Evidence encounter

How long before the evidence was encountered did the event occur?

Clock 2: Evidence encounter → Report

How long before the report was the evidence encountered?

The two intervals are not necessarily equal.

An event may have happened recently, while the evidence was encountered much later.

Or an event may have happened long before the evidence was discovered.

The discovery itself may also be recent or distant relative to the moment of speaking.

Matsés provides grammatical resources for expressing such relationships.

This is why the term double tense is more informative than simply saying that the language has several past tenses.

17. Why the order of knowledge matters

Suppose someone says:

I discovered yesterday that the building had collapsed ten years ago.

The statement includes two different pieces of temporal information.

The collapse happened ten years ago.

The discovery happened yesterday.

Now compare:

I discovered ten years ago that the building had collapsed shortly before that.

The same general event type is involved.

But the speaker’s relationship to the information is different.

The timing of discovery matters.

It affects how the information entered the speaker’s knowledge.

Matsés makes this relationship grammatically salient in inferential past-event reporting.

The language therefore offers a striking example of how time and information source can be integrated into one grammatical system.

18. Recent, distant and remote past

Fleck describes several temporal distinctions within the Matsés evidential system.

For experiential past, the principal suffixes include:

For inferential past, relevant forms include:

The temporal boundaries are approximate and sensitive to the grammatical and discourse context.

They should not be treated as a universal calendar with perfectly fixed cut-off dates.

The important point is that the evidential categories interact with distinctions of temporal distance.

The grammar does not merely ask:

Was it seen or inferred?

It also asks how the relevant event and evidence are situated in time.

19. Conjecture is a separate category

Matsés also has forms associated with conjecture.

Among them are:

-ash

and:

-nëdash

These represent conjectural past categories with different temporal interpretations.

A conjecture is not equivalent to an inference grounded in concrete consequences of the event.

Imagine seeing fresh footprints and concluding that an animal passed.

That is an evidence-based inference.

Now imagine believing that an animal probably passed because animals often use that path, without having observed fresh traces.

That is a different kind of judgment.

The distinction is not always easy to reproduce through a single English word.

Expressions such as:

apparently

probably

must have

I suppose

can overlap in everyday use.

But their grammatical and pragmatic functions are not identical to the Matsés categories.

20. The problem with “apparently”

English apparently is a useful example of translation difficulty.

Consider:

Apparently, he left.

What does the speaker mean?

Perhaps someone told them.

Perhaps they found his room empty.

Perhaps they saw his luggage disappear.

Perhaps they have other reasons for drawing the conclusion.

The English adverb may leave the precise source underdetermined.

That is not necessarily a weakness.

English allows the speaker to omit distinctions that are irrelevant to the immediate communicative task.

But when translating Matsés, a single English adverb may fail to preserve the specific evidential category encoded in the original.

The translator must decide whether the information source is important enough to express explicitly.

21. Hearsay works differently

Now consider:

Someone told me that he left.

This is neither direct experience of the departure nor necessarily an inference from physical evidence.

The speaker has received a report.

In Matsés, hearsay is expressed through a quotation-based construction.

This is an important grammatical distinction.

Matsés does not simply add a reportative suffix parallel to all its experiential and inferential suffixes.

Instead, a speaker can use a construction involving reported speech and a verb of saying.

This means the source category of hearsay is expressed through a different grammatical mechanism.

The functional system of information-source distinctions is therefore broader than one uniform set of suffixes.

22. A documented Matsés quotation

Research on reported speech in Matsés gives an example involving a speaker named Dashe.

The structure is represented as:

Dashe [kachina pe-o-mbi] ke-o-sh

With the meaning:

“I ate chicken,” said Dashe.

The important feature is that the original speaker’s perspective is retained inside the quotation.

The reported statement is not simply transformed into an ordinary English-style indirect clause.

The quotation-based construction plays a central role in how Matsés represents information obtained from another person’s speech.

This is particularly important because the hearsay mechanism is linked to the language’s broader system of evidential reporting.

23. Reported speech is not always a literal recording

Here we need another scientific correction.

A quotation-based construction does not necessarily mean that every word must be reproduced exactly as originally spoken.

Research by Robert Munro, Rainer Ludwig, Uli Sauerland and David W. Fleck shows that Matsés reported speech has a distinctive combination of properties.

Personal, spatial and temporal expressions retain the perspective of the original speaker.

Yet paraphrasing and certain kinds of restructuring are possible.

This makes the system especially interesting.

It resembles direct speech in its handling of perspective.

But it also allows operations commonly associated with indirect speech.

The result cannot be explained simply by saying:

Matsés speakers always quote word for word.

The grammar is more subtle than that.

24. Who is “I” in reported speech?

Imagine Maria says:

I was at home yesterday.

Later, Peter reports her statement.

English offers two possibilities.

Direct speech:

Maria said, “I was at home yesterday.”

Indirect speech:

Maria said that she had been at home the previous day.

Notice what changes.

I becomes she.

Yesterday becomes the previous day.

The perspective shifts from Maria’s original speaking situation to Peter’s reporting situation.

Matsés reported speech follows a different pattern in its handling of these indexical expressions.

The original speaker’s perspective is preserved.

This is not merely a stylistic preference.

It is a grammatical property of the reported-speech system.

25. The relationship between Matsés and Ewe

In our earlier investigation, He Said That He Left — But Whose “He”? What Ewe Reveals About Grammar and Another Person’s Point of View, we examined logophoric pronouns.

Ewe provides grammatical resources for identifying a participant in relation to the perspective of a person whose speech or thought is being represented.

Matsés raises a related but different question.

How does the grammar of reported speech preserve the original speaker’s personal, spatial and temporal perspective?

And how does that reporting system interact with evidentiality?

The two languages do not use identical mechanisms.

But both demonstrate that reported speech is not simply the mechanical insertion of someone else’s sentence into a new sentence.

Perspective itself can be grammatically organized.

26. Why reported information is not automatically false

A common misunderstanding is that hearsay means unreliable information.

But information obtained from another person may be accurate.

A scientist may report a measurement performed by a colleague.

A doctor may receive laboratory results.

A journalist may cite a verified statement.

A witness may relay an event accurately.

The information source is indirect, but the content may be correct.

Conversely, direct observation can be mistaken.

Therefore, a language that distinguishes hearsay from direct experience is not necessarily ranking every statement according to truth.

It is distinguishing the speaker’s relationship to the information.

Reliability requires additional evaluation.

27. What if the speaker lies?

An obligatory evidential system does not make deception impossible.

A speaker can misrepresent the source of information.

They can claim to have experienced something they did not experience.

They can conceal the fact that a conclusion is speculative.

They can repeat false information received from another person.

The grammatical system creates expectations about how information should be presented.

It does not independently investigate whether those expectations have been satisfied.

Fleck’s work discusses the social consequences of choosing an inappropriate evidential construction.

But this must not be exaggerated into the claim that Matsés grammar automatically prevents lying.

No grammatical suffix can replace verification.

28. The social consequences of source marking

When a language routinely requires information-source distinctions, speakers and listeners can develop expectations about appropriate reporting.

If someone presents an inferred event as directly experienced, the statement may be treated as inaccurate or deceptive.

If information comes from another speaker, a quotation-based construction can identify that relationship.

The grammatical choice therefore has consequences beyond sentence structure.

It participates in how speakers present responsibility for their claims.

However, we should distinguish between a grammatical convention and a universal cultural judgment.

We cannot infer from the existence of evidential suffixes that all Matsés speakers are more honest, more skeptical or more scientifically minded than speakers of other languages.

Those claims would require independent evidence.

29. Grammar can require a distinction without guaranteeing good reasoning

Consider two people.

The first says:

I saw the animal.

The second says:

I inferred that the animal was present from its tracks.

The source distinction is explicit.

But now ask:

Were the tracks correctly identified?

Could another animal have made them?

How old were the tracks?

Was the observation reliable?

The source classification does not settle these questions.

It tells us what kind of information relationship is being claimed.

The evaluation of evidence is a separate task.

This distinction is essential for understanding the philosophical significance of Matsés.

30. The difference between evidence and interpretation

Evidence does not speak by itself.

Someone must interpret it.

Imagine finding broken glass.

You conclude:

Someone broke the window.

But perhaps the glass was broken by a falling branch.

Or pressure.

Or an accident.

The physical trace exists.

The interpretation may be wrong.

Therefore, even an evidence-based inference contains at least two layers.

Observation: broken glass is present.

Interpretation: someone broke the window.

A grammar can mark the statement as inferential.

But it does not automatically identify every assumption involved in the reasoning.

The listener still needs to distinguish observed facts from conclusions drawn from them.

31. The same event can produce several kinds of knowledge

Imagine a storm damages a bridge.

Four people learn about it differently.

One person sees the bridge collapse.

Another arrives later and observes the damage.

A third hears the news from a neighbour.

A fourth assumes the bridge probably collapsed because the storm was severe.

They may all eventually state:

The bridge collapsed.

But their informational positions are different.

The first has direct experience.

The second has evidence-based inference.

The third has a report.

The fourth has conjecture.

The proposition is the same.

The route to that proposition is different.

This is the central insight of evidentiality.

32. What English can do — and what it does not require

English is fully capable of expressing all these distinctions.

I watched the bridge collapse.

The damage suggests that the bridge collapsed.

A neighbour told me the bridge collapsed.

I suspect the bridge collapsed.

The difference lies in grammatical obligatoriness and organization.

English usually allows the speaker to omit the information source.

Matsés normally requires an appropriate source-related construction when reporting past events.

That does not mean English speakers are incapable of distinguishing evidence from speculation.

Nor does it mean Matsés speakers cannot reason about uncertainty beyond their grammatical categories.

It means the languages distribute communicative obligations differently.

33. The grammar of evidence and the grammar of time

Our earlier study of Yagua and temporal distance examined how a language can encode distinctions concerning how far into the past an event occurred.

Matsés adds another dimension.

The language does not merely distinguish temporal distance.

It connects temporal distinctions to the source of information.

And in inferential constructions, it can encode two relevant temporal relationships.

When did the event occur relative to the evidence encounter?

When was the evidence encountered relative to the report?

This is why Matsés is not simply another example of a language with several past tenses.

Its distinctive contribution lies in the integration of time and evidence.

34. Why this matters for journalism

Imagine a headline:

The company concealed financial losses.

What does the journalist actually know?

Did they examine internal documents?

Did an employee make an allegation?

Did an audit establish the discrepancy?

Did the journalist infer concealment from published accounts?

The headline presents a conclusion.

But the source and quality of the supporting evidence matter.

English journalism can distinguish these situations through careful attribution.

For example:

Internal documents show…

A former employee alleges…

An independent audit found…

The available records suggest…

These expressions are not equivalent.

Matsés does not provide a universal model for journalistic ethics.

But its grammatical system offers a powerful comparative reminder:

A claim and the grounds for making it are different pieces of information.

35. Why this matters for medicine

Consider a medical report.

The patient experienced dizziness.

Was the dizziness observed directly by the clinician?

Reported by the patient?

Recorded by another healthcare professional?

Inferred from a related symptom?

The distinction matters.

A clinician may write:

The patient reports dizziness.

That identifies the source.

Or:

Examination revealed…

That identifies a finding from the examination.

Or:

The findings are consistent with…

That expresses an interpretation.

The clinical language does not become more accurate simply by sounding more confident.

It becomes more useful when observations, reports and interpretations are appropriately distinguished.

The parallel with evidentiality is conceptual rather than a claim that medical English has Matsés-style evidential morphology.

36. Why this matters for translation

Suppose a Matsés statement contains an inferential marker.

An English translator may render it as:

He left.

That may preserve the basic event.

But it may lose the fact that the speaker inferred the departure rather than witnessing it.

Depending on context, a fuller translation may be necessary:

He apparently left.

It seems he left.

The evidence suggests he left.

But these expressions are not perfect equivalents in every context.

The translator must consider what kind of inference is encoded and what the English expression implies.

The same problem arises with conjecture and reported speech.

Translation requires reconstructing the relationship between the speaker and the information, not merely transferring the event description.

37. The translator must not invent certainty

Imagine a source statement expresses conjecture.

The translator renders it as a categorical fact.

The event description may remain unchanged.

But the informational status has been altered.

Conversely, imagine a source statement presents direct experience.

The translator adds:

perhaps

without justification.

Now the target text suggests uncertainty that may not have been present in the source.

Both transformations can misrepresent the original.

A responsible translation must distinguish:

what happened;

what the speaker claims happened;

how the speaker knows;

and how strongly the speaker evaluates the conclusion.

These dimensions are not always encoded in the same grammatical locations across languages.

38. Understanding is not the same as agreement

There is a practical communication problem that goes beyond evidentiality.

Someone says:

I understand.

What exactly has happened?

They may have heard the statement.

They may have understood its words.

They may have understood the intended action.

They may agree with it.

Or they may disagree while understanding perfectly.

These are different states.

Likewise, a person can understand the source of a claim without accepting the claim as true.

I understand that you inferred this from the evidence does not mean I agree that your inference is correct.

The distinction matters in professional communication, education and everyday decision-making.

Grammar can help represent these relationships.

But speakers must still reason about them.

39. How this relates to language learning

A student learning English may encounter:

must have

might have

apparently

allegedly

reportedly

seem to

according to

These expressions cannot all be taught as interchangeable ways of saying:

maybe.

They differ in meaning and use.

Consider:

He must have forgotten.

The speaker presents a strong inference.

He might have forgotten.

The speaker presents a possibility.

He reportedly forgot.

The speaker attributes the information to a report.

He apparently forgot.

The interpretation may depend on context and the grounds available to the speaker.

Understanding these differences requires more than memorizing translations.

It requires recognizing the communicative operation performed by each construction.

40. The danger of teaching “must” as certainty

English learners are often told:

must = strong certainty

But this is incomplete.

Compare:

You must leave now.

An obligation.

He must be at home.

An inference or strong conclusion.

He must have left.

An inference about a past situation.

The form participates in different modal constructions.

And the inference does not necessarily mean the speaker directly witnessed the event.

A learner who understands the relationship between evidence, inference and modality will interpret these constructions more accurately than someone who memorizes a single translation.

Matsés gives us a useful comparative perspective on why these distinctions matter.

41. Does Matsés make its speakers more rational?

This is a tempting but unsupported conclusion.

A language with obligatory evidential distinctions may encourage speakers to express information sources explicitly in certain contexts.

But that does not prove that its speakers are more rational than speakers of English.

Rationality involves many factors.

The quality of evidence.

The validity of inference.

The willingness to revise beliefs.

The ability to detect contradictions.

The social conditions of communication.

The existence of evidential morphology does not establish superiority in any of these dimensions.

The responsible linguistic conclusion is narrower.

Matsés makes certain source-of-information distinctions grammatically central to past-event reporting.

That is important without requiring an exaggerated psychological claim.

42. What does cognitive research tell us?

Research on evidentiality and cognition has examined whether grammatical source marking influences how people process or remember information.

Such studies suggest that evidential distinctions can interact with linguistic interpretation and cognitive tasks.

However, the relationship between language and cognition is complex.

A grammatical distinction does not automatically create a cognitive ability.

Nor does the absence of obligatory grammatical marking mean speakers cannot recognize the distinction.

English speakers can distinguish firsthand observation from hearsay.

Matsés speakers can evaluate whether an inference is reliable.

The languages differ in how those relationships are conventionally encoded.

We should therefore investigate cognitive effects empirically rather than assume them from grammatical structure alone.

43. Why this is not just another version of Turkish -miş

Turkish provides a well-known example of evidential or indirect-information marking, particularly through forms associated with -miş.

But Matsés offers a substantially different research problem.

The Matsés system includes a detailed interaction between evidential categories and temporal distinctions.

Its inferential constructions can encode two temporal relationships.

Its hearsay reporting relies on a quotation-based construction rather than a parallel reportative suffix.

Its reported-speech system also preserves distinctive perspective relations.

Therefore, the value of Matsés is not simply that it has a grammatical way to say:

I heard that…

Its contribution lies in how source, time, inference and reported perspective interact.

44. The difference between information and knowledge

A person can receive information without knowing whether it is true.

A person can observe evidence without interpreting it correctly.

A person can make a valid inference from reliable premises.

A person can also make a confident mistake.

These distinctions are essential.

Information source concerns where the content came from.

Evidence concerns what supports a conclusion.

Inference concerns the reasoning that connects evidence to a claim.

Knowledge raises further questions about truth and justification.

Matsés evidentiality does not solve all these philosophical problems.

But it makes one important component of the problem unusually visible in grammar.

The speaker’s relationship to the information is not simply left outside the sentence.

It becomes part of the grammatical choices involved in reporting past events.

45. What the Matsés evidence actually establishes

We can now summarize the linguistic findings without exaggeration.

Matsés has a highly elaborate system for distinguishing sources of information in past-event reporting.

Experiential, inferential and conjectural categories are expressed through specialized verbal morphology.

Hearsay is handled through a quotation-based construction.

The system interacts with temporal distinctions.

Certain inferential forms encode both the temporal relationship between an event and the evidence encounter, and the relationship between the evidence encounter and the present report.

Reported speech preserves important aspects of the original speaker’s perspective.

These are significant grammatical facts.

They do not prove that every Matsés statement is true.

They do not prove that Matsés speakers are more honest.

They do not prove that English speakers cannot distinguish evidence from speculation.

They show that languages can assign very different grammatical roles to the sources and timing of information.

Conclusion: the grammar of a claim is not the proof of a claim

We began with a simple sentence:

Someone entered the house.

The event could have been witnessed.

Inferred from physical traces.

Reported by another person.

Or merely suspected.

The proposition might remain the same.

But the speaker’s relationship to it changes.

Matsés demonstrates that this relationship can be deeply integrated into grammar.

Its evidential system distinguishes direct experience, inference and conjecture, while hearsay is handled through reported-speech constructions.

Its double-tense system reveals an even more intricate possibility.

The grammar can distinguish when an event happened from when the evidence for that event was encountered.

That means language can organize not only events in time, but also the speaker’s acquisition of information about those events.

Yet the most important lesson is not that grammatical evidence marking guarantees truth.

It does not.

The lesson is that an assertion contains more than its surface proposition.

There is also a relationship between the speaker, the event and the grounds for making the claim.

English can express that relationship.

Matsés makes important parts of it grammatically obligatory.

And the comparison forces us to ask a question that matters far beyond linguistics:

When someone tells us what happened, do we know how they know?

Knowing what someone claims is only the beginning. Understanding the grounds for that claim is where evaluation becomes possible.

— Tymur Levitin

Linguistic sources and further reading

1. David W. Fleck (2007). Evidentiality and Double Tense in Matses. Language, 83(3), 589–614. Research article. The principal source for Matsés evidential suffixes, past-time distinctions, inferential constructions and double tense.
2. Robert Munro, Rainer Ludwig, Uli Sauerland and David W. Fleck (2012). Reported Speech in Matses: Perspective Persistence and Evidential Narratives. International Journal of American Linguistics, 78(1). Research article. A detailed analysis of quotation, perspective and evidential reporting.
3. Alexandra Y. Aikhenvald (2004). Evidentiality. Oxford University Press. A major comparative study of grammatical information-source marking across languages.
4. Alexandra Y. Aikhenvald (2014). The Grammar of Knowledge: A Cross-Linguistic View. In The Grammar of Knowledge: A Cross-Linguistic Typology. Oxford University Press. Publisher preview.
5. Anna Papafragou, Peggy Li, Youngon Choi and Chung-hye Han (2007). Evidentiality in Language and Cognition. Cognition, 103(2), 253–299. Open-access research article. Research on the relationship between linguistic evidential distinctions and cognition.

About the author and language learning

Tymur Levitin is the founder and director of Levitin Language School, an international online school providing individual language education.

His educational approach emphasizes understanding linguistic mechanisms, distinguishing meaning from form and developing the ability to communicate in real situations.

This article belongs to the Language as a Way of Thinking research series, which explores how languages organize experience, evidence, perspective, time and grammatical meaning.

The series is a comparative linguistic research project. It does not imply that every language examined is offered as a course.

For international online language learning, visit Language Learnings.

For professional translation, editing, proofreading, copywriting and multilingual adaptation, explore Professional Language Services.

For individual online lessons or enquiries, contact Tymur Levitin via Telegram.

Tymur Levitin
Founder & Director, Levitin Language School
Language Teacher, Translator and Author

© Tymur Levitin / Levitin Language School. All rights reserved.