A sentence does not only tell us what happened. Sometimes its grammar prepares us for who will carry the story forward.
Consider two tiny stories.
I opened the door and walked outside.
I opened the door and Peter walked inside.
English handles them easily.
In the first sentence, the person who opens the door is also the person who walks outside.
In the second, the participant changes.
Nothing particularly remarkable seems to happen grammatically.
English gives us:
and
and then relies heavily on subjects, pronouns, context and our general ability to follow a narrative.
But imagine a language in which the first event itself can carry grammatical information about what happens at the boundary with the next clause:
the next relevant subject will be the same
or
the next relevant subject will be different.
Before the second event has fully unfolded, the grammar has already helped organize the transition.
This phenomenon is known as switch-reference.
And Amele, a Papuan language of Papua New Guinea, gives us an exceptionally clear window into it.
The important question is not:
What strange suffix does Amele use?
It is:
Why can the identity of the participant in the next event become part of the grammar of the transition from this event?
1. English makes us reconstruct continuity
Take:
Sarah entered the room and sat down.
English speakers immediately understand that Sarah sat down.
Now:
Sarah entered the room and John sat down.
The subject changes.
Again, there is no difficulty.
But notice how the information is distributed.
In the first sentence, English allows the second subject to remain unexpressed:
Sarah entered the room and Ø sat down.
The listener carries Sarah forward.
In the second:
Sarah entered the room and John sat down.
the overt noun tells us that the participant has changed.
English therefore certainly tracks reference.
Every language must somehow allow people to understand who is doing what.
But languages need not package that information in the same place.
Amele can grammaticalize the relationship between clauses themselves.
2. Same subject or different subject?
The classic switch-reference distinction is usually described with two abbreviations:
SS — same subject
DS — different subject
The basic idea is wonderfully simple.
Suppose we have:
CLAUSE 1 → CLAUSE 2
A marker associated with the first clause can tell us whether its subject has the same reference as the subject of the following clause.
So conceptually:
I did X — SS → I did Y
versus:
I did X — DS → he/she/they did Y
The first event does not merely end.
Its grammatical form helps specify how the participant structure continues into the next event.
That is the essential mechanism.
3. The marker looks forward
This is what makes switch-reference particularly interesting.
We often imagine grammatical marking as describing the clause we are currently processing.
Past tense tells us about its temporal relation.
Number tells us something about a participant.
Case can tell us something about grammatical or semantic relations.
But switch-reference creates a special kind of bridge.
A marker on one clause can tell us something about the referential relationship between that clause and another clause.
In canonical cases, the listener receives an instruction roughly like:
keep following the same subject
or:
prepare for a subject switch.
Grammar is therefore helping manage continuity across a boundary.
4. A sentence can contain a prediction about its own continuation
Not prediction in the sense of guessing the future.
The speaker already knows what they intend to say.
But for the listener, the morphology can constrain expectations.
Imagine hearing:
I went to the river — SAME SUBJECT…
Before the next lexical verb is fully processed, one piece of information is already available:
the relevant participant continues.
Now:
I went to the river — DIFFERENT SUBJECT…
The listener knows that the next clause will not simply continue with the same subject in the canonical pattern.
The grammar has reduced the search space.
That is a remarkable communicative function.
5. English has continuity too — but distributes it differently
This does not mean English is unable to express the distinction.
Compare:
After eating, John left.
After John ate, Mary left.
John ate and then left.
John ate, and then Mary left.
English uses:
pronouns,
repeated nouns,
zero subjects in certain coordinated structures,
subordination,
word order,
discourse context,
and ordinary inference.
The information is available.
What differs is its grammatical address.
Amele has a dedicated switch-reference system that makes the relation between subjects across linked clauses morphologically relevant.
Once again, our central question is not:
Can English say it?
Almost always, English can communicate the event.
The question is:
What does the other language require its grammar to keep track of while saying it?
6. Amele is particularly interesting because clauses can form chains
Amele makes extensive use of linked clauses in which non-final, often called medial, verbs depend in important ways on the final clause.
This means a narrative can develop as a sequence of events before reaching the fully finite endpoint of the chain.
The medial verbs can carry switch-reference marking.
So instead of treating every event as a completely self-contained sentence, the grammar can build something more like:
event → event → event → event → endpoint
And along that route, it can help track whether the relevant subject continues or changes.
The sentence becomes not merely a container.
It becomes a path.
7. Who carries the event chain?
Imagine:
The boy woke up, got dressed, left the house, walked to the river and started fishing.
English repeats the subject only once.
We infer continuity.
Now introduce another participant:
The boy woke up, got dressed, left the house, and his father called him.
At the final transition, the subject changes.
In an Amele-style switch-reference system, those transitions are not all structurally equivalent.
Some maintain participant continuity.
Another breaks it.
The grammar can distinguish them.
The story therefore has not only events.
It has a referential trajectory.
8. Reference is not the same thing as repetition
This distinction matters.
Two clauses can contain no repeated noun while still referring to the same participant.
And two clauses can contain similar expressions while referring to different participants.
Switch-reference is fundamentally about reference, not merely whether the same word appears twice.
Consider:
The teacher entered and she sat down.
The expressions:
the teacher
and
she
are different.
The referent may be the same.
Conversely:
A teacher entered and another teacher sat down.
The noun is repeated.
The referents differ.
Grammar must ultimately interact with the world of discourse, not merely compare strings of words.
9. Same subject does not mean “same noun”
This is why the label same subject can mislead beginners.
It does not mean:
the same grammatical expression was repeated.
It means, in the canonical case, that the relevant subjects are coreferential.
They point to the same discourse participant.
So:
Tymur arrived and he sat down
contains two different expressions:
Tymur
he
but one referent.
A switch-reference system is interested in that identity relation.
It is tracking who, not merely which word.
10. Different subject does not necessarily mean a new character
The opposite misunderstanding is equally important.
A DS marker does not mean:
a brand-new person has entered the story.
Suppose Peter and Maria have already been mentioned.
Then:
Peter spoke — DS → Maria answered.
Maria is not new.
The switch concerns the subject relation across the relevant clause boundary.
This is a local grammatical distinction, not a literary instruction saying:
introduce new character here.
That difference will become important when we move from textbook examples to real discourse.
11. Why not just use pronouns?
An English speaker might reasonably ask:
Why does a language need switch-reference when pronouns already tell us who is acting?
But this assumes every language must solve reference tracking through the same division of labour.
Pronouns identify or help recover participants.
Switch-reference does something related but structurally different.
It marks a relationship between reference points across clauses.
Compare the information:
he
versus:
same as the preceding subject
They are not identical instructions.
One points toward a participant.
The other describes continuity between grammatical units.
12. Think of a relay race
A useful model is a baton.
Each clause carries an event.
At the boundary, grammar can tell us:
the same runner keeps the baton
or:
the baton passes to another runner.
The baton is not literally the subject.
And real switch-reference is more complicated than this metaphor.
But the model reveals something useful:
the important information lies not only inside each clause.
It also lies in the transition between clauses.
Traditional grammar teaching often underestimates transitions.
Amele makes them impossible to ignore.
13. Grammar can describe connections, not just things
Students are often taught grammar as a list of properties:
noun number,
verb tense,
case,
person,
gender,
aspect.
These categories seem to belong to individual words or clauses.
Switch-reference reminds us that grammar can also encode relations between larger units.
Clause A does not simply have its own structure.
Its form can depend on how Clause B relates to it.
This means grammatical meaning can live partly at the boundary.
That is a much more dynamic picture of language.
14. And Amele makes the story more complicated
If switch-reference were always:
same grammatical subject = SS
different grammatical subject = DS
we could stop here.
But natural language rarely respects our desire for perfectly clean classroom boxes.
Research on Amele has shown cases in which the morphology cannot be reduced mechanically to literal subject identity.
The system interacts with broader properties of discourse and event organization.
Changes involving time, place or world setting can also matter in the distribution of switch-reference morphology.
That is where the phenomenon becomes much more interesting.
15. What exactly is continuing?
At first we thought the answer was:
the subject.
Now consider a broader possibility.
Perhaps discourse continuity involves several dimensions:
participant continuity
temporal continuity
spatial continuity
event continuity
world or setting continuity
A narrative can preserve one while changing another.
The same person can continue acting but:
move to another place,
enter another time,
move into a hypothetical world,
begin a structurally distinct event sequence.
Conversely, a participant can change while a larger event remains tightly integrated.
So the grammar–discourse interface becomes richer than a binary identity test.
16. This does not make SS and DS meaningless
We should not overcorrect.
Canonical switch-reference really does centrally concern referential identity between subjects or subject-like privileged arguments.
Modern grammatical descriptions of Amele continue to characterize the system in precisely those terms.
The noncanonical cases do not mean:
SS and DS actually have nothing to do with subjects.
They mean:
the simple rule captures the prototype, not every discourse condition governing real usage.
This distinction is crucial.
A pedagogical model can be correct without being complete.
17. A rule can be true and still not be the whole mechanism
This principle extends far beyond Amele.
We teach:
SS = same subject
DS = different subject
That is useful.
Then real data reveal exceptions or extensions.
A poor response is:
The rule was wrong.
Another poor response is:
Ignore the inconvenient examples.
The better response is:
The rule described the central contrast. Now we need a larger model explaining why the system behaves this way outside the prototype.
That is how understanding develops.
Not by abandoning rules.
By discovering their domain.
18. This is exactly why grammar should not be taught as commandments
Consider the difference between:
Use this marker when the subject stays the same.
and:
This system helps organize continuity and discontinuity across linked clauses; subject coreference is its central grammatical contrast.
The first statement is easier to memorize.
The second gives the learner somewhere to put unusual cases later.
That difference matters enormously.
A memorized rule breaks when reality becomes more complicated.
A model can expand.
19. What does the listener gain?
Imagine listening to a long sequence of events involving several people.
Every new clause creates potential uncertainty:
Who is acting now?
Did the previous participant continue?
Did another participant take over?
Does this event belong to the same chain?
Has the narrative moved somewhere else?
Languages possess many devices for managing these questions.
Switch-reference is one of them.
By marking continuity or discontinuity at clause boundaries, the grammar can help the listener maintain a structured representation of the unfolding discourse.
It is not decorative morphology.
It participates in information management.
20. The listener is always predicting
Human language processing does not wait passively until a sentence is finished.
Listeners continually form expectations.
After:
If I had known…
you expect a certain kind of continuation.
After:
Although…
you anticipate a contrast.
After:
Either…
you expect an alternative.
After a name followed by:
he
you search for a plausible referent.
Switch-reference adds another type of cue.
It can tell the listener something about the upcoming referential relationship before all the content of the following clause is available.
Grammar is helping cognition stay ahead of the acoustic stream.
21. Language disappears as soon as we hear it
Speech has a brutal property.
It vanishes.
A written sentence remains on the page.
Spoken language arrives through time and disappears behind us.
So listeners need mechanisms for maintaining structure while new information keeps arriving.
Who was acting?
What happened first?
Which participant does this expression refer to?
Are we still inside the same event sequence?
Switch-reference can contribute to solving exactly this kind of problem.
It helps turn a transient stream into an organized discourse model.
22. Compare this with Pirahã
Our previous investigation of Pirahã asked how symbolic language can help preserve exact cardinality after the original objects disappear.
Amele takes us somewhere different, but there is an intriguing structural parallel.
In Pirahã:
the objects disappear, but information may need to survive.
In Amele:
the clause disappears into the past of the utterance, but participant structure must survive into the next clause.
These are not the same phenomenon.
But both remind us that language is not only about naming what is immediately present.
Language also helps maintain continuity across disappearing information.
23. Guugu Yimithirr asked where the coordinates come from
In Guugu Yimithirr, we asked where a spatial coordinate system begins.
Does it rotate with the observer?
Or remain anchored to a larger environment?
Amele introduces another kind of coordinate:
discourse reference.
Who is the current participant?
Does that reference point persist?
Does it switch?
Again, language is managing relationships rather than merely attaching labels to objects.
The mechanisms are completely different.
The deeper lesson is converging.
24. Translation can preserve the story and erase the tracking system
Suppose an Amele clause chain corresponds naturally to:
He got up, went outside, saw his brother, and called him.
English can tell the story perfectly well.
The listener can follow it.
Nothing essential about the events has to be lost.
But the grammatical architecture can disappear.
English may express participant continuity through:
zero repetition,
pronouns,
lexical subjects,
coordination,
and inference.
The source language may have explicitly marked relationships between adjacent clauses through switch-reference morphology.
The translation preserves:
what happened.
It may not preserve:
how the original grammar helped the listener track who continued doing it.
25. This is becoming a recurring translation problem
Yagua showed us that translation can preserve when an event happened while moving temporal remoteness from grammar into vocabulary.
Guugu Yimithirr showed us that a translation can produce the correct spatial instruction while changing the underlying frame of reference.
Amele now shows another possibility:
translation can preserve the participant sequence while removing the grammatical machinery that tracks continuity between clauses.
The information survives.
Its architecture changes.
This may be one of the most important general principles in comparative linguistics.
26. Translation is not only about what survives
We often judge translation by asking:
Did the target sentence preserve the meaning?
That remains essential.
But comparative analysis asks another question:
What had to be structurally encoded in the source in order to produce that meaning?
Two sentences can describe the same event and impose different grammatical decisions on their speakers.
That difference matters even when communication succeeds perfectly.
A translation may therefore be semantically excellent while still hiding the mechanism that made the source sentence possible.
27. English speakers already understand the underlying problem
Switch-reference may sound exotic.
The communicative problem is not.
Consider:
John told Peter that he had to leave.
Who had to leave?
John?
Peter?
English pronouns can create ambiguity.
We resolve such problems using:
syntax,
context,
world knowledge,
prosody,
discourse expectations.
So English speakers already know what reference tracking feels like.
Amele simply demonstrates that languages can grammaticalize aspects of this task differently.
The exoticism disappears as soon as we identify the problem being solved.
28. The sentence is not the whole unit of thought
Traditional teaching often treats the sentence as the natural endpoint of grammar.
Capital letter.
Words.
Full stop.
Finished.
But real language unfolds across:
clauses,
sentences,
turns,
topics,
episodes,
narratives.
Participants persist across these boundaries.
Events connect.
Reference shifts.
Time moves.
Settings change.
Switch-reference belongs precisely to this larger architecture.
It forces us to look beyond the isolated sentence.
29. Perhaps grammar is partly traffic control
A useful metaphor emerges.
Discourse contains moving streams:
participants
events
times
places
reference points
Grammar helps prevent collisions.
Pronouns redirect us.
Conjunctions connect routes.
Tense locates events.
Aspect tells us how they unfold.
Switch-reference signals continuity and change across clause boundaries.
Grammar is not merely a collection of forms.
It is partly a system for controlling the flow of information.
30. This changes what “knowing grammar” means
Suppose a student memorizes:
SS = same subject
DS = different subject
and correctly fills ten gaps in an exercise.
Do they understand switch-reference?
Perhaps at the first level.
But deeper understanding requires different questions:
What exactly is being compared?
Why is the marker located where it is?
What information does it provide to the listener?
What happens in clause chains?
Why do some real examples depart from literal subject identity?
What kinds of discourse continuity interact with the system?
How would English express the same information?
What disappears in translation?
Now grammar becomes a mechanism rather than a label.
31. Thinking Instead of Memorizing means asking what problem the grammar solves
This does not mean every grammatical form exists because someone deliberately designed it to solve a communicative problem.
Languages evolve historically.
Forms change.
Constructions are reanalysed.
Systems accumulate irregularities.
But synchronically we can still ask:
What information is this distinction organizing for speakers and listeners?
With Amele, one answer is clear.
The grammar helps monitor reference across linked clauses.
And once we understand that, SS and DS stop being arbitrary abbreviations.
They become two possible instructions for navigating discourse.
32. The next event begins before the previous clause is forgotten
Speech moves only forward.
We cannot keep every preceding word equally active in consciousness.
So successful communication depends on compression.
We retain:
who matters,
what happened,
what continues,
what changed.
Switch-reference is one grammatical contribution to that compression.
Instead of forcing the listener to reconstruct every participant relationship from scratch, the language can mark a transition.
Continue this referent.
Switch away from it.
The story keeps moving.
33. And sometimes the story itself switches
This is where the noncanonical uses become philosophically interesting.
A participant can remain the same while something larger changes.
Time.
Place.
Setting.
Discourse world.
The grammar can become sensitive to discontinuity beyond a simple change of person.
This suggests that human discourse is organized not merely as:
person A → person B → person C
but as structured episodes.
An episode has participants.
A location.
A temporal frame.
A world.
A coherent event structure.
Switch-reference systems can interact with those boundaries.
34. “Same” is therefore not always a simple idea
Same person.
Same subject.
Same topic.
Same event.
Same time.
Same place.
Same discourse world.
These are different kinds of sameness.
Languages may align them in typical cases.
Real discourse can pull them apart.
And when that happens, grammar reveals which type of continuity matters to the system.
This is why unusual examples are valuable.
They show us where our first definition stops working.
35. The exception can reveal the mechanism
If every Amele example mechanically followed:
same subject → SS
different subject → DS
we could describe the distribution.
But deviations force a deeper question:
What is the system actually tracking?
That question is one of the most productive questions in linguistics.
An exception is not automatically noise.
Sometimes it is the place where the hidden architecture becomes visible.
Conclusion: grammar can tell us how the story crosses a boundary
A person acts.
Then another event begins.
Perhaps the same person continues.
Perhaps someone else takes over.
Perhaps the participant remains but the setting changes.
Perhaps an entire discourse episode shifts.
Every narrative contains thousands of such transitions.
English usually manages them through a mixture of subjects, pronouns, conjunctions, syntax, context and inference.
Amele shows that a language can make continuity across clauses a dedicated part of its grammatical architecture.
At its canonical core, switch-reference asks an elegantly simple question:
Is the relevant subject of the next clause the same or different?
But that simple distinction opens onto something larger.
Language does not merely describe isolated events.
It has to connect events while speech is disappearing behind us.
It must preserve participants.
Signal changes.
Manage expectations.
Build coherence.
And sometimes grammar marks the transition before the listener has finished discovering what comes next.
The story does not simply consist of events.
The story also consists of who carries us from one event into another.
Grammar is not only the structure inside an utterance. Sometimes it is the bridge that tells us what is allowed to continue across its next boundary.
Tymur Levitin
Founder & Director, Levitin Language School
Certified Translator and Language Teacher
© Tymur Levitin / Levitin Language School. All rights reserved.
