Sometimes grammar does not begin by asking who is the subject. It begins by asking who stands higher in the conversation.
English gives us two extremely ordinary sentences:
You see me.
I see you.
Nothing mysterious seems to be happening.
The verb see stays see. The participants change places around it.
you → see → me
becomes:
I → see → you
If we want to know who sees whom, English gives us several familiar clues: pronoun form, syntactic position and word order.
Now look at two forms from Plains Cree, or nêhiyawêwin:
ki-wâpam-i-n
“You see me.”
and:
ki-wâpam-iti-n
“I see you.”
Something remarkable has happened.
Both forms begin with ki-.
Yet the direction of seeing has reversed.
The crucial difference is inside the verb:
-i-
versus:
-iti-
This is our entrance into a grammatical architecture that can feel profoundly unfamiliar if we approach it through English.
It is usually described as a direct–inverse system.
But the most interesting question is not what the terminology means.
It is this:
Why should a language need to tell us that an action is moving in a “direct” or “inverse” direction at all?
1. First forget the English sentence
The easiest way to misunderstand Plains Cree is to begin with:
Which word means I?
Which word means you?
Which word is the subject?
Which word is the object?
Those questions are not meaningless.
But they encourage us to reconstruct Cree as if it were English with different vocabulary.
Instead, start with the participants.
There are two people:
I
and
you.
One sees the other.
The event can run in either direction:
YOU → ME
or:
ME → YOU
English changes the arrangement of independent words.
Plains Cree can encode the relationship through the internal morphology of the verb.
The verb does not merely name SEEING.
It helps construct a map of who acts upon whom.
2. But the participants are not grammatically neutral
Here comes the stranger part.
In the traditional analysis of the Plains Cree direct–inverse system, participants occupy positions in a hierarchy commonly represented approximately as:
2 > 1 > 3 > 3′
That is:
second person > first person > third-person proximate > third-person obviative
So you is ranked above I for this grammatical system.
This is not a ranking of human worth.
It does not mean Cree speakers believe that “you are more important than me.”
It is a grammatical hierarchy used in the organization of person marking and direction.
And once the hierarchy exists, the language can ask another question about a transitive event:
Does the action travel in the expected direction of the hierarchy, or against it?
3. You see me: direct
Consider:
ki-wâpam-i-n
“You see me.”
Schematically:
2 → 1
The second person acts upon the first.
Because the direction follows the relevant hierarchy:
2 > 1
the construction uses direct morphology.
We can simplify the architecture as:
YOU → ME = DIRECT
Now reverse reality.
I see you.
The same two participants are present.
The same lexical event — seeing — is present.
But the arrow has changed direction.
4. I see you: inverse
Now we have:
ki-wâpam-iti-n
“I see you.”
Schematically:
1 → 2
The first person acts upon the second.
But the hierarchy still ranks:
2 > 1
So the semantic action is travelling against that hierarchy.
The verb marks the relationship as inverse.
Schematically:
ME → YOU = INVERSE
This is one of those moments when translation can conceal almost everything interesting.
English gives us:
You see me.
I see you.
Cree gives us not merely different participants in different slots, but a grammatical indication of the direction of the relationship between them.
5. “Inverse” does not mean “backwards”
The terminology is dangerous.
The moment a learner hears inverse, it is tempting to imagine:
normal action → reversed action.
But nothing in the event itself is backwards.
If I see you, I simply see you.
The world is perfectly ordinary.
What is “inverse” is the relation between the semantic roles and the grammatical participant hierarchy.
That distinction matters.
Grammar is not telling us:
Something strange happened.
It is telling us, approximately:
The participant lower in the relevant hierarchy is acting upon the participant higher in it.
The event has not been inverted.
Its grammatical alignment with the hierarchy has.
6. Now add a third person
The system becomes even more revealing when first or second person interacts with third person.
Compare the well-described Plains Cree pattern:
ni-wâpam-â-w
“I see him/her.”
and:
ni-wâpam-ik
“He/she sees me.”
Again, notice what survives.
The first-person marking remains associated with the verb.
But the direction morphology tells us whether the first person acts on the third person or the third person acts on the first.
Schematically:
1 → 3 = DIRECT
but:
3 → 1 = INVERSE
The important question is no longer merely:
Which noun came first?
It becomes:
Which participant acts on which participant, and how does that relation fit the hierarchy?
7. This is not simply the passive voice
An English speaker may immediately think:
So inverse is basically passive.
No.
That shortcut is attractive precisely because it makes the unfamiliar system look familiar.
But it destroys the mechanism we are trying to understand.
Compare:
I see him.
I am seen by him.
English passive changes the syntactic organization of the clause and promotes what would otherwise be the object into subject position.
The Plains Cree inverse system belongs to a different grammatical architecture.
Inverse morphology is systematically tied to relationships among participants on a hierarchy.
Indeed, with first and second person, the inverse form is not some stylistically marked alternative chosen to avoid an active sentence.
For the relevant relationship, it is part of the ordinary grammatical system.
Calling it “passive” may therefore help for approximately five seconds — and mislead the learner for considerably longer.
8. And third person is not just “third person”
Now the system opens another door.
Suppose a discourse contains two animate third-person participants.
English can keep saying:
he
she
the man
the woman
and rely heavily on lexical repetition, syntax, pronouns and context to keep reference clear.
Plains Cree has another grammatical distinction:
proximate versus obviative.
Very roughly, when multiple animate third persons are present, one can occupy the more discourse-prominent proximate position while another is marked as obviative.
The hierarchy can therefore be extended:
2 > 1 > 3 > 3′
where 3′ represents an obviative third person.
Now direct and inverse morphology can operate between third persons too.
The grammar is not merely tracking:
person 1
and:
person 2.
It is also tracking their relative discourse status.
9. “He sees him” is suddenly not a trivial sentence
English gives us a wonderfully ambiguous little sentence:
He sees him.
Outside context, we know almost nothing about the identities of the two men.
But imagine a discourse in which one third-person participant is already central and another is less central.
Plains Cree grammar can distinguish these participants through the proximate–obviative system and can then mark whether the action travels:
proximate → obviative
or:
obviative → proximate.
The first relation is associated with direct morphology.
The reverse relation calls for inverse morphology.
This means that grammatical direction is interacting not only with semantic roles such as agent and patient, but with the way participants are organized inside the discourse.
And that gives us a much deeper insight.
A sentence is not always built from a flat list of nouns.
Sometimes grammar is already carrying a map of attention.
10. But “proximate = important” would also be too crude
We should resist another seductive simplification.
It would be easy to write:
Cree grammar marks the important person and the unimportant person.
That is not accurate enough.
Proximate and obviative are grammatical and discourse categories.
Topicality and discourse prominence help us understand them, but neither English word important nor a psychological notion of “who matters more” is an adequate definition.
A person can be morally central but grammatically obviative in a particular stretch of discourse.
Another participant can occupy proximate status because of how that discourse is structured.
Once again:
working model ≠ entire linguistic reality.
The model helps us see the mechanism.
It should not replace the mechanism.
11. Why is “you” above “I”?
This is perhaps the most irresistible question in the hierarchy:
2 > 1
Why should second person outrank first person?
One answer is historical and grammatical: this is simply how the person hierarchy is organized in the Algonquian system.
But cognitively, the contrast is fascinating because it reminds us that grammatical person is not merely a numbering system.
“First person”, “second person” and “third person” sound as if someone simply counted:
- me
- you
- somebody else
But those labels conceal different communicative roles.
I am the current speaker.
You are the current addressee.
He/she is outside that speaker–addressee pair.
The speech situation therefore already has an internal geometry.
Plains Cree makes grammatical use of that geometry in ways English generally does not.
12. The verb becomes a small model of the event
For an English learner, a verb can initially appear to be something like:
SEE + tense
see
saw
have seen
will see
But a Plains Cree transitive verb can require us to think in a richer way.
The form may encode information connected with:
- the lexical event,
- the persons involved,
- their number,
- animacy,
- direction,
- and, in relevant contexts, proximate–obviative relations.
The verb is therefore not simply the word in the middle of:
PERSON — VERB — PERSON.
It can behave more like a compact grammatical representation of the participant structure itself.
13. Word order loses its English monopoly
English strongly trains us to trust order:
The dog sees the man.
is different from:
The man sees the dog.
Exchange the noun phrases and the event changes.
Plains Cree does not rely on a fixed English-style word order to do all of this work.
Its morphology carries information that English often assigns to syntactic position.
That does not mean Cree word order is meaningless or random.
Flexible word order is not free word order.
Ordering can interact with discourse, emphasis and information structure.
But the division of grammatical labour is different.
English asks word order to do a great deal.
Cree morphology does a great deal of work elsewhere.
14. Translation preserves the event and changes the architecture
This is where the phenomenon joins several other investigations in our growing map of language.
When we examined Basque ergative–absolutive alignment, the same real-world event could be organized through a participant alignment unfamiliar from English.
When we examined Fijian possession, English my collapsed distinctions that Fijian grammar could make explicit.
Plains Cree now gives us a third problem.
Translation can preserve:
who saw whom
while making almost invisible the fact that the source language has grammatically encoded the action as direct or inverse relative to a participant hierarchy.
Nothing necessarily disappears from the basic event.
But something disappears from the architecture by which the event was built.
That distinction is fundamental.
Same event does not mean same linguistic model of the event.
15. So does Cree make its speakers think hierarchically?
No such conclusion follows automatically.
This is precisely where interesting linguistics can turn into bad popular science.
From the existence of a grammatical hierarchy, we cannot jump to:
Cree speakers see human beings as a hierarchy.
Nor can we conclude:
Cree speakers perceive interpersonal relationships more accurately.
Nor:
English speakers cannot perceive these distinctions because English does not grammaticalize them in the same way.
Humans can understand distinctions their language does not obligatorily encode.
English speakers certainly understand that discourse has central and peripheral participants.
They understand that I see you and you see me reverse participant roles.
The significant fact is narrower — and stronger because it is defensible:
Plains Cree grammar requires speakers to encode some participant relationships through a grammatical system that English does not possess.
That alone is remarkable.
We do not need linguistic determinism to make it interesting.
16. What should a learner memorize?
The conventional answer might begin with a paradigm.
Learn:
direct
inverse
proximate
obviative
and then memorize the endings.
But terminology is not yet understanding.
A better first model is visual.
Put two participants on the page.
Then draw the hierarchy:
2 > 1 > 3 > 3′
Now draw an arrow representing the event.
If the actor is higher and the affected participant lower:
→ DIRECT
If the action runs against the hierarchy:
← INVERSE
Only after the learner understands what the grammar is solving should the morphology be attached to the model.
Then:
ki-wâpam-i-n
and:
ki-wâpam-iti-n
are no longer two arbitrary strings to memorize.
They are two solutions to two different participant configurations.
That is the difference between memorizing a paradigm and understanding why a paradigm exists.
17. But even the hierarchy is only a model
And now we have to challenge our own explanation.
Writing:
2 > 1 > 3 > 3′
is extremely useful.
But a hierarchy on a page can tempt us to believe that we have reduced the language to a formula.
We have not.
Actual Cree grammar involves different verb classes, animacy, person and number paradigms, independent and conjunct orders, obviation, discourse structure and morphological patterns far richer than a four-position diagram.
The hierarchy is a map.
It is not the territory.
The point of a good model is not to eliminate complexity.
It is to tell us where the complexity comes from.
18. A completely different question about “subject” and “object”
After Basque, we already had reason to distrust the assumption that every language must package grammatical relations exactly like English.
Plains Cree pushes the problem from another direction.
Instead of asking only:
Who is the subject?
and:
Who is the object?
we may also need to ask:
Who are the participants?
Where do they stand in the grammatical hierarchy?
Which one acts upon the other?
Does the event follow that hierarchy or run against it?
Which third person is discourse-prominent?
These are not decorative questions added after the sentence has been built.
They participate in determining the form of the verb itself.
Conclusion: grammar can encode an arrow
English makes I see you and you see me look like a simple exchange of pronouns.
Plains Cree reveals that the same conceptual reversal can be organized through a radically different grammatical mechanism.
The language can establish a hierarchy among grammatical persons and discourse participants.
The verb can then indicate whether an action follows that hierarchy or goes against it.
So the event contains more than:
SEE
and two people.
It contains an arrow.
And the grammar knows which way that arrow points.
This does not prove that Cree speakers inhabit a different reality.
It demonstrates something more precise and, in my view, more interesting:
human languages do not merely give different names to the same pieces of reality. They can choose different relations as the ones grammar must make visible.
Sometimes another language does not change who did what. It changes what the sentence must tell you before you can know who did what.
Tymur Levitin
Founder & Director, Levitin Language School
Certified Translator and Language Teacher
© Tymur Levitin / Levitin Language School. All rights reserved.
