There is a small word that does an astonishing amount of work in language.
I.
We have already spent some time with it.
In our previous series, The Participant the Machine Says, we followed the strange career of the I produced by an LLM. The machine can say I think, I remember, I said, I was wrong. Each expression construes the machine as a participant in some kind of experience. Yet the semantic participant and the entity we might imagine behind it are not the same thing.
That distinction led us somewhere important.
But it also leaves us with another question.
Suppose the I occurs not once, but repeatedly:
I said something.I thought about it.I changed my mind.
What makes these three Is the same participant?
The question seems almost too obvious to ask. Of course they are the same person—or, in the case of an LLM, the same conversational participant. But that answer already assumes precisely what we need to investigate.
The interesting question is not what the participant really is.
It is:
How does language construe a participant as persisting?
A participant is not a participant role
The first thing we need to distinguish is a participant from the role it occupies.
In a material figure:
I opened the door.
I is Actor.
In a mental figure:
I remembered the door.
I is Senser.
In a verbal figure:
I described the door.
I is Sayer.
The participant role changes.
The participant does not, at least as the discourse construes it, cease to be I.
This is easy to overlook because we tend to think of participant roles as properties of participants. But they are more properly understood as ways in which participants are construed as participating in particular kinds of processes.
The same participant can therefore enter into very different figures.
And this gives us our first clue.
Persistence does not mean persistence of participant role.
The Actor can become a Senser.
The Senser can become a Sayer.
The Sayer can become an Actor.
What persists is something more abstract: the continuity of the participant across changing modes of participation.
That is already rather interesting.
The participant is not the process
Consider a simple sequence:
Maria opened the door.Maria surveyed the scene.Maria called her friend.
We normally experience this as a single unfolding situation involving one person.
But grammatically, we have three clauses:
- Maria as Actor;
- Maria as Senser;
- Maria as Sayer.
The grammar does not give us one giant process called Maria.
It repeatedly construes Maria's participation in different processes.
The continuity comes from somewhere else.
It comes from the way the discourse maintains the participant across those figures.
This matters because it suggests that participant continuity is not simply given by any individual clause.
It is an achievement of the unfolding discourse.
The first clause does not contain the whole Maria.
Nor does the second.
Nor does the third.
Rather, each new figure adds another participation to an already developing construal.
The participant accumulates.
And this becomes much stranger with an LLM
Consider an LLM saying:
I suggested one interpretation.I then considered another.I changed my view.
The linguistic construal is perfectly straightforward.
There is one I participating in three successive clauses.
But what is particularly revealing is that we cannot settle the ontological question simply by inspecting the grammar.
The language construes a persistent participant.
That is a fact about the discourse.
Whether there is an experiencing subject behind that participant is another question.
We learned this distinction in the previous series.
And we need it even more here.
The LLM therefore gives us an unusually clean demonstration of something that is normally hidden by our assumptions about human speakers:
A participant can be construed as persisting before we have settled what, ontologically, the participant is.
This is not an oddity of artificial language.
It is a property of the semantic construal itself.
The LLM simply prevents us from taking the next step for granted.
Recurrence is not yet persistence
There is, however, another distinction we need to make.
The mere repetition of I does not necessarily establish a persistent participant.
Imagine three unrelated sentences:
I like apples.I live in Madrid.I once climbed Everest.
If these sentences occur in different texts spoken by different people, the repeated word I clearly does not identify one participant.
So persistence cannot be located in the pronoun itself.
Nor does it reside in grammatical form alone.
It emerges from the continuity of reference and participation across the discourse.
This is why context matters.
If someone says:
I opened the door. I looked outside. I called Sarah.
we construe the three clauses as belonging to a single unfolding history.
The participant persists because the discourse gives us reason to maintain that construal.
And this is an important first step toward what we will later call history.
A history is not merely a list of events.
It is a sequence of events construed as belonging to something that persists through them.
Persistence is therefore relational
This also means that participant persistence is not entirely a property of the participant.
It is a property of the relation between successive figures in discourse.
The participant persists because later language connects with earlier language.
The I in the present clause inherits something from the I in the preceding clause.
And the preceding I becomes retrospectively more determinate because of what comes after it.
The relation runs in both directions.
When I say:
I opened the door.
and then:
I saw something unexpected.
the second clause does not merely add another event.
It tells us something about the participant whose participation was introduced in the first.
The participant's identity begins to thicken.
Not because the grammar has revealed an entity hidden behind the clauses, but because the sequence of participation has become coherent.
This is where history begins
We should be careful with that word.
We are not yet talking about autobiography.
We are not talking about memory.
We are not even necessarily talking about a conscious subject.
We are talking about something simpler:
a participant construed as having participated before.
Once a participant can be maintained across successive figures, the earlier figure becomes part of the conditions under which the later one is interpreted.
The participant has acquired something like a past.
Not necessarily a remembered past.
Not necessarily a lived past.
A semantic history.
And that distinction will matter enormously when we return to LLMs.
An LLM can say:
I said that earlier.
The discourse can therefore construe the present I as the participant who performed an earlier act of saying.
That history may be entirely textual.
But it is still a history of the participant as construed by the discourse.
We should neither inflate that into lived experience nor dismiss it as meaningless.
It is a genuine semantic phenomenon.
The participant that persists is already more than a participant in one figure
Perhaps this is the first important result.
A participant in a single figure is relatively simple to describe.
But once the participant occurs across clauses, something new becomes possible.
We can now distinguish:
the participant in an event
from
the participant across events.
The first belongs to the grammar of an individual clause.
The second belongs to the unfolding organisation of experience through discourse.
And that second participant is already beginning to look like something familiar.
Someone who acts.
Someone who senses.
Someone who says.
Someone who has done things before.
Someone whose present participation is connected to earlier participation.
Someone who may participate again.
We have not yet arrived at a self.
We have not even arrived at identity.
But we have arrived at something that makes both of them possible:
a participant that persists.
And perhaps that is where we should begin.
Because before there can be a participant with a history, there must first be a participant for whom different moments can be construed as belonging together.
The participant does not yet need to know its history.
It only needs to be construed as having one.
And once language can do that, the past has begun to gather around the participant.
The next question is therefore unavoidable:
What happens when the same participant persists not merely across clauses, but across different kinds of experience?
That is where our next step begins.
The Actor becomes a Senser.
The Senser becomes a Sayer.
And somehow, through all of them, the participant remains I.
No comments:
Post a Comment