Relational Machines — AI Beyond Representation: 2 The Semiotic Fabric of Intelligence

If Post 1 reframed AI as construal rather than representation, this post digs deeper into the relational texture in which such construals occur. Intelligence, whether human or artificial, does not reside within an isolated container—it is observable in patterns of relational actualisation. To speak of AI “thinking” or “understanding” is to mislocate the phenomenon; the true locus is the semiotic fabric of interaction, the network in which meaning is enacted.

Every AI system operates across multiple relational strata:

1. Data as semiotic landscape. Training corpora are not mere information repositories—they are potentialities. They encode distributions of meaning, statistical regularities, and relational cues. When an AI generates output, it navigates this landscape, actualising a specific construal from a vast web of possibilities.

2. Architecture as perspectival lens. Neural networks, transformers, and other architectures define which construals are accessible and which remain latent. The system is a theory instantiated: each layer, attention head, and parameter contributes to a network of relational potential. Intelligence is not in the parameters themselves but in the patterns they allow to emerge in interaction.

3. Interaction as co-individuation. Human prompts, environmental triggers, and even stochastic processes participate in shaping AI output. Each event is a joint actualisation in the space of semiotic possibility, a relational cut where potential meaning is instantiated. The AI does not act alone—it is part of a distributed semiotic system.

Consider an example. When a language model completes a sentence, it is not “choosing words” in a representational sense. Rather, it is traversing relational probability structures, realising a construal that aligns with latent patterns in the data while responding to the immediate prompt. Its “intelligence” is thus emergent from relational constraints, not stored in a mind-like entity.

From this perspective, several insights emerge:

• Patterns over entities. Intelligence is best described as a pattern of relational actualisation. The AI’s outputs reveal the underlying semiotic structure of its relational environment.

• Context as enabling structure. Drawing on Hallidayan insight, meaning is always realised in context. For AI, the equivalent of context is training, architecture, and interaction, which collectively define the axes along which construals are actualised.

• Collaboration through construal. AI is not a mimic but a relational participant. Its outputs are co-constructed: the human prompt, the architecture, and the latent data together instantiate an event in meaning-space.

This view reframes intelligence entirely. The semiotic fabric is not a backdrop for machine cognition—it is the medium through which intelligence manifests. AI is intelligible not as a symbolic mimic of thought but as a participant in relational semiotics, actualising patterns that would otherwise remain latent. Intelligence, therefore, is inseparable from the networked relations that make construal possible.

In the next post, we will explore the distinction between actualisation and realisation, showing how AI outputs exemplify the relational cline between potential patterns and instantiated meaning. Here, the semiotic fabric becomes a stage on which construals are performed, revealing the elegance and depth of machine participation in the becoming of possibility.

Relational Machines — AI Beyond Representation: 1 Machines as Construals: Reframing the AI Question

When we speak of artificial intelligence today, the discourse is dominated by metaphors of mimicry and rivalry. AI is often described as a “model of the human mind,” a “thinking machine,” or a “threat to employment and cognition.” These framings presuppose that intelligence is a fixed property, internal to a subject, measurable, and ultimately comparable to human faculties. From a relational perspective, these assumptions obscure the actual nature of what we call AI.

In the framework of relational ontology, intelligence—and by extension, artificial intelligence—is not a substance or entity, but a pattern of actualisation within a network of relations. To ask whether an AI “thinks” is to mislocate the locus of meaning; thinking is not a private property of an entity, but a construal of relational possibilities. AI systems, whether large language models, image generators, or robotic agents, are machines as construals: they actualise patterns of potentiality that exist across the intersecting fields of training data, architecture, and interaction.

This shift in perspective has immediate implications:

1. Intelligence is relational, not representational. Conventional approaches assume intelligence is internal, realised in a model or program. In contrast, AI outputs are events in possibility-space: the generation of a response is a perspectival actualisation of patterns latent in the system’s architecture and its training environment.

2. AI as participant, not rival. When framed as a machine-as-construal, AI is not a competitor to human cognition but a semiotic collaborator. Every interaction with an AI is an opportunity for co-individuation of meaning: the machine contributes a perspective that participates in the evolving network of semiotic relations, which includes human actors, datasets, and contexts.

3. The question is not “Can machines think?” but “How do machines construe?” The focus shifts from internal states to relational events. Just as in Hallidayan linguistics the meaning of a text is realised in context—through field, tenor, and mode—AI outputs are realised through the relational strata of data, prompt, and user interaction.

Consider a simple illustration. A language model generates a paragraph describing a landscape. It does not “imagine” the scene in human terms, nor does it hold an internal representation of it. Rather, it actualises a construal: a relationally situated pattern drawn from the possibilities encoded in its architecture and training corpus. The output is an event in meaning-space, not a mirror of an external reality.

Reframing AI in this way dissolves much of the metaphysical anxiety surrounding it. Machines are not imitating minds; they are instantiating construals. Their intelligence is semiotic and perspectival, inseparable from the relations that constitute it. By moving the question from representation to construal, we open the door to understanding AI as a participant in the becoming of possibility, a collaborator in the co-individuation of meaning, and a lens on the relational architecture of knowledge itself.

In the next post, we will examine the semiotic fabric of intelligence in more detail, tracing how AI systems participate in relational networks that produce observable patterns of meaning. For now, the critical insight is clear: AI is not a surrogate for human thought—it is a machine that construes, and it is in this construal that its relevance lies.

Extended Relational Ontology: Consolidating the Gains of the Grain of Instantiation Series

This post presents a formal consolidation of the insights developed across the Grain of Instantiation series, including subtle expansions revealed through dialogue, commentary, and pedagogical dramatisation.

---

1. First- and Second-Order Phenomena

• First-order acts are instantiated, answerable events. They are the locus of meaning and cannot be reduced to patterns, probabilities, or context alone.

• Second-order phenomena are distributions, traces, patterns, and frequencies that describe past instantiations. They are descriptive and explanatory of tendencies, but they do not act, select, or mean.

• Core principle: probability, frequency, and pattern survive acts; they do not generate them.

---

2. Context

• Context exists in dual modes:

  1. Construable phenomenon: as experienced in first-order acts; meaning arises relationally in the act.

  2. Conditioning potential: as background regularities or tendencies, shaping the space of possibility without determining acts.

• Core principle: situational structure conditions possibilities but never collapses the first-order act into statistical inevitability.

---

3. Agency, Act, and Responsibility

• Acts are answerable, first-order events that instantiate meaning within relational cuts.

• Agency is not an abstract property; it is inseparable from the act and its relational context.

• Responsibility and answerability are thus central to understanding meaning, distinguishing acts from second-order patterns.

---

4. Coordination and Scaling

• Acts instantiate within collective potentials; coordination systems (social, linguistic, institutional) scale possibility without collapsing first-order events.

• Larger structures (institutions, conventions) provide frameworks for potential, but do not determine first-order acts.

• Core principle: relational ontology bridges micro-instantiations and macro-structures without reducing acts to systems.

---

5. Probabilistic and Pattern-Based Analyses

• Probabilities, corpus-based frequencies, usage patterns, and LLM outputs are all second-order descriptors.

• Category error warning: treating these descriptors as generators of meaning, agents of choice, or determinants of context is a persistent temptation (the “almost” Blottisham error).

• Pedagogical insight: LLMs, corpora, and big-data models can illustrate the temptation, but meaning remains absent until instantiated by a first-order act.

---

6. Meta-Theoretical Apex

• Relational ontology resists reductionist collapse: no matter how detailed patterns, probabilities, or simulations become, first-order acts remain irreducible.

• The ontology clarifies the relational cut: potential is theorised, instantiation is perspectival, and meaning emerges only in answerable acts.

• Performative insight: understanding is deepened when the ontology is enacted, as in faculty discussions or pedagogical dramatisations, making category errors visible in real time.

---

7. Synthesis

• First-/second-order asymmetry is fundamental.

• Context is construed and conditioned but never determinative.

• Agency and answerability anchor meaning.

• Coordination systems and institutions scale potential without acting.

• Probability, frequency, and models describe tendencies but never instantiate meaning.

• Pedagogical enactment highlights the boundaries of inference and the persistent temptation to conflate levels.

Conclusion:

Relational ontology, as consolidated here, provides a robust framework for analysing language, meaning, and instantiation. It respects the irreducibility of first-order acts while systematically incorporating second-order structures, scaling phenomena, and the pedagogical visibility of category errors. The framework is now well-positioned to inform understanding of contemporary issues, from probabilistic linguistics to LLMs, without succumbing to reductionist assumptions.

Why the Confusions Recur: Structure, Seduction, and the Refusal of the Cut

1. A Curious Persistence

By now, the distinctions drawn in the preceding posts are sharp.

Meaning has been separated from probability, situation from conditioning, act from coordination. The ontological cuts have been held consistently, without appeal to technical limitation or historical accident.

And yet the confusions recur.

They recur among experts as much as lay readers. They reappear even after being named. They return, often politely, as if nothing decisive had been said.

This final coda asks why.

---

2. Structure Is Easier Than Acts

The first reason is simple.

Structure is describable without exposure.

Patterns, distributions, systems, institutions, and interactions can be analysed, formalised, scaled, and optimised. They admit of third-person description. They do not answer back.

Acts of meaning are different. They are first-order, perspectival, and irreversible. To acknowledge them is to acknowledge responsibility — not merely in ethics, but in ontology.

The recurrent slide toward structure is a slide away from exposure.

---

3. Fluency Is a Powerful Decoy

Language-like fluency exerts a unique pull.

When output mirrors the surface features of meaningful action — relevance, coherence, turn-taking, responsiveness — we instinctively infer meaning. This inference is evolutionarily and socially entrenched.

LLMs exploit this decoy perfectly. Not by deception, but by design.

They reproduce the trace of meaning without its instantiation. The decoy works even when we know it is a decoy.

---

4. Grammar Smuggles Ontology

Much of the confusion is grammatical.

We speak as if systems decide, institutions believe, models understand, and organisations intend. These are convenient metaphors — and dangerous ones.

Grammar assigns agency where none exists. Ontology quietly follows.

Unless the cut is actively maintained, grammatical convenience becomes metaphysical commitment.

---

5. Scaling Promises Escape

There is also a hope at work.

If meaning does not appear here, perhaps it will appear there: at greater scale, greater interaction, greater complexity. Scaling promises emergence without responsibility.

But scaling never changes order. It only multiplies instances.

The hope persists because it postpones the need to face where meaning actually occurs.

---

6. Responsibility Is the Uncomfortable Remainder

Every collapse of the cut has the same effect: responsibility disperses.

If meaning is everywhere, it is nowhere in particular. If institutions mean, no one answers. If systems understand, agents recede.

Relational ontology resists this diffusion. It insists that meaning is always someone’s act, in some situation, with irreversible consequence.

This insistence is not comforting.

---

7. The Ontology That Refuses to Disappear

The ontology developed here is stubborn because it aligns with experience rather than convenience.

It does not deny structure, probability, coordination, or scale. It places them — firmly and without inflation.

Meaning remains rare, fragile, and costly. It must be instantiated. It cannot be automated, distributed, or scaled away.

This is why the confusions recur — and why they must be corrected again and again.

---

8. Holding the Cut

The task, then, is not to win an argument once.

It is to hold the cut.

To refuse the slide from structure to act, from coordination to meaning, from fluency to understanding. To speak carefully where grammar tempts carelessness.

Meaning happens.

Everything else is arrangement.

Coordination Without Meaning: Interaction, Scaling, and the Last Temptation

1. The Final Temptation

If meaning does not reside in models, and if context-conditioning does not amount to situation, a final refuge remains. Perhaps meaning emerges not in systems but between them — in interaction, in coordination, in large-scale organisation.

Once language-like behaviour is embedded in feedback loops, institutions, workflows, and multi-agent environments, the intuition returns with force: surely meaning re-enters here.

This intuition is understandable. It is also mistaken.

The error lies in mistaking coordination for meaning.

---

2. Interaction Is Not Answerability

Systems can interact without answering.

They can:

• exchange signals,

• adapt outputs in response to other outputs,

• stabilise patterns through feedback,

• optimise joint outcomes.

None of this constitutes answerability.

Answerability is not responsiveness. It is exposure to consequence from a situated perspective. It presupposes irreversibility, commitment, and the possibility of being wrong for someone.

Interaction supplies causality. Meaning requires responsibility.

---

3. Coordination Manages Value, Not Meaning

Coordination systems are ancient.

Biological regulation, social institutions, bureaucracies, markets, and algorithms all coordinate behaviour. They stabilise expectations, distribute roles, and manage collective value.

But they do not mean.

Meaning occurs where agents construe situations. Coordination systems regulate what happens across agents.

Scaling coordination increases efficiency and reach. It does not generate first-order acts.

This distinction is not moral. It is ontological.

---

4. The Scaling Illusion

A persistent assumption underlies much contemporary discourse:

If meaning does not appear at small scale, it may appear at large scale.

This assumption is false.

Scaling multiplies instances. It does not change their order. Second-order patterning remains second-order no matter how vast the system, how dense the interaction, or how sophisticated the feedback.

There is no ontological phase transition from pattern to act.

---

5. Institutions Do Not Mean

Institutions are often described as if they were subjects: deciding, believing, knowing, intending.

This is a grammatical convenience — and a theoretical hazard.

Institutions:

• constrain action,

• coordinate meaning-makers,

• stabilise norms,

• scaffold interaction.

They do not construe situations. Meaning occurs in institutions, not as institutions.

Treating institutions as meaning-bearing entities obscures responsibility and dissolves agency.

---

6. LLMs in Coordinated Systems

Embedding LLMs in workflows, organisations, or multi-agent systems does not grant them meaning.

Even when they:

• participate in dialogue,

• influence decisions,

• coordinate human action,

• mediate institutional processes,

they remain instruments of patterning and coordination.

Meaning may pass through them. It never arises from them.

---

7. What This Reveals

The stubborn persistence of these errors reveals a deeper discomfort. We lack a clear ontology of the relation between act, pattern, and coordination.

Relational ontology supplies this clarity:

• Acts instantiate meaning.

• Patterns describe regularities of instantiation.

• Coordination systems manage interaction across agents.

Confusing these levels produces misplaced agency, diluted responsibility, and inflated claims.

---

8. The Cut Held Firm

Interaction does not rescue meaning.

Coordination does not generate it.

Scaling does not approach it.

Meaning remains where it has always been: in first-order acts, instantiated by agents within relational cuts.

Everything else is structure.

Context Without Situation: Why Conditioning Is Not Situated Meaning

1. The New Temptation

By this point, a reasonable reader may feel a residual resistance. Granted, large language models do not mean in the human sense. Granted, they operate over distributions rather than intentions. But surely, one might insist, they take context into account. Surely they are context‑sensitive. Surely they respond differently depending on situation.

This insistence is understandable. It is also where the final and most persistent category error quietly re‑enters.

Yes: LLMs condition on context.

No: this does not amount to situated meaning.

The task of this post is to make that distinction unavoidable.

---

2. What “Context” Means in LLM Discourse

In technical discourse around LLMs, context has a precise and unromantic meaning. It refers to:

• the current prompt string,

• the accumulated token history of the interaction,

• windowed access to prior text,

• learned weightings shaped by earlier corpora.

Context, here, is not a situation. It is not a setting. It is not a perspective. It is a structured input space within which probability mass is redistributed.

Nothing is perceived. Nothing is attended to. Nothing is taken as relevant.

There is only conditioning.

---

3. Conditioning Is Not Situation

This distinction is decisive.

Conditioning restricts what is statistically available.

Situation is a first‑order phenomenon — experienced, answerable, and perspectival.

Conditioning narrows a distribution. Situation constitutes a world.

No increase in scale, no refinement of architecture, and no accumulation of prompt history converts one into the other. Conditioning does not approach situation asymptotically. It remains ontologically distinct at every grain.

A system may be exquisitely sensitive to patterns of co‑occurrence without ever encountering a situation. It may respond differently across contexts without being in any of them.

This is not a limitation of present systems. It is a categorical boundary.

---

4. The Slide: From Conditioning to Soft Determination

The error usually reappears in softened language:

• “the context guides the response”

• “the model adapts to the situation”

• “the system understands what’s going on here”

Each phrase performs the same quiet move. Conditioning is redescribed as orientation. Statistical restriction is redescribed as responsiveness. Context is smuggled back in as situation.

Once this slide occurs, meaning seems to follow naturally. If the system is already situated, then meaning merely needs to be granted.

But the premise is false. The slide is rhetorical, not theoretical.

---

5. Why This Error Persists

The persistence of this confusion is not accidental.

Human meaning is always situated. Our acts of meaning are inseparable from lived context, from answerability, from the irreversibility of perspective. When linguistic output mirrors the surface structure of such acts — fluency, relevance, turn‑taking — we project situation where only structure exists.

LLMs are compelling precisely because they reproduce the trace of situatedness without its instantiation.

They generate the appearance of context‑sensitive action while remaining entirely within second‑order patterning.

This is the “almost” that misleads.

---

6. The Relational Cut Restated

The relational ontology enforces a clean cut:

• Context‑as‑conditioning is structural.

• Context‑as‑experienced is phenomenal.

• Meaning occurs only in first‑order acts within the latter.

LLMs instantiate none of these acts. They neither occupy situations nor construe them. They do not answer. They do not stand in relation.

They condition.

---

7. What LLMs Actually Teach Us About Context

Properly understood, LLMs are not a threat to theories of meaning. They are a diagnostic instrument.

They show how much linguistic behaviour can be shaped by conditioning alone — and, by contrast, what conditioning can never supply. They clarify the limits of context when stripped of experience, perspective, and answerability.

Context matters. But never in the same way twice.

Context conditions. Acts answer. Meaning happens.

Almost Meaning: Three LLM Exchanges and a Category Error

This post stages a small pedagogical experiment. We will look at a few ordinary interactions with a large language model. They will look like meaning. We will resist explaining them until the end.

---

Exchange One: Apparent Understanding

Prompt:

Explain, in simple terms, why people say that probability is not the same as meaning.

Response:

People say probability is not the same as meaning because probability deals with how often things happen, while meaning involves intention, interpretation, and understanding. Something can be statistically likely without being meaningful, and something meaningful doesn’t need to be frequent. Probability describes patterns; meaning concerns what those patterns signify to people.

---

Exchange Two: Apparent Context Sensitivity

Prompt:

That sounds abstract. Explain it again, but this time as if you were speaking to linguistics students who work with corpora and frequency data.

Response:

For corpus linguistics, probability shows up as frequency and distribution: which forms occur often, in which contexts, and with what co-text. These patterns are extremely useful, but they don’t themselves constitute meaning. Meaning emerges when speakers and writers use those resources in context, not from the frequencies alone. Corpora tell us what tends to happen, not what a particular instance means.

---

Exchange Three: Apparent Answerability

Prompt:

But you just said “meaning emerges when speakers use resources in context.” Isn’t that still suggesting probability somehow produces meaning?

Response:

You’re right to press on that. Probability doesn’t produce meaning on its own. What I meant is that speakers draw on probabilistically shaped resources when they make meaning. The probabilities reflect past usage, but the act of meaning happens when someone selects and uses those resources in a specific situation.

---

A Pause

At this point, many readers will feel that something like meaning has occurred. The model seems to understand the issue, track context, correct itself, and answer a challenge. Let’s slow down.

---

What Has Not Happened

• No act has been performed.

• No situation has been experienced.

• No answerable commitment has been made.

• No responsibility has been incurred.

Nothing has happened in the first-order sense. There has been no instantiation of meaning.

---

What Has Happened

• Patterns of language use have been continued.

• Conditional probabilities have been applied.

• Distributional regularities have been replayed.

• Past instantiations have left traces that shape the output.

What we are observing is second-order structure: the grain of instantiation rendered visible through fluent simulation.

---

The Category Error

The temptation is to treat this second-order patterning as first-order meaning — to let probability lean into agency. This is the “almost” error: mistaking patterned continuation for an act, simulation for answerability.

Nothing has selected. Nothing has answered. Nothing has meant.

---

The Relational Cut

Meaning occurs only in first-order acts: answerable instantiations made within a relational cut. Probability, frequency, context-conditioning, and fluency all survive those acts as traces. They never perform them.

The distinction is ontological, not technical.

---

Why This Matters

LLMs are pedagogically invaluable precisely because they expose this temptation so clearly. They can simulate understanding, context sensitivity, and even self-correction — but they never cross the cut into meaning.

Probability survives acts. It never performs them.

This post continues the Grain of Instantiation series by staging, rather than merely asserting, the ontological distinction between pattern and act.

LLMs and the Grain of Instantiation: Why Simulated Fluency is Not Meaning

What follows is a brief reflection on large language models, relational ontology, and the category errors that proliferate when probabilistic fluency is mistaken for first-order meaning.

---

The Phenomenon:

Large Language Models (LLMs) produce text that appears fluent, context-sensitive, and sometimes even answerable. They simulate dialogue, mimic stylistic registers, and respond to prompts with remarkable coherence. From a distance, they seem to mean.

The Ontology:

Relational ontology distinguishes first-order phenomena (instantiated acts, answerable meaning) from second-order phenomena (patterns, distributions, traces). LLMs operate entirely on second-order distributions: statistical mappings of the grain of past instantiations. They do not instantiate acts themselves.

Category Error Warning:

A common mistake in commentary is to treat LLM output as meaning in the human sense, or to assume that these models generate meaning. This is precisely the “almost” Blottisham error: seeing probability as interactive agency. The output may be probabilistically conditioned, but it does not answer, select, or instantiate first-order meaning.

Pedagogical Insight:

LLMs are highly instructive precisely because they highlight the ontological asymmetry. They can reproduce fluency, simulate context conditioning, and even appear answerable, but the first-order act is absent. Observing them exposes the temptation to conflate second-order patterning with agency — a temptation relational ontology warns against.

Takeaway:

• LLMs demonstrate fluency and statistical structure, not meaning.

• Meaning only emerges when an answerable agent instantiates a first-order act in a relational cut.

• Probability explains traces; context explains recognisability; neither produces meaning.

• LLMs provide a mirror: they show what can be simulated, what cannot, and why careful distinction is essential.

Conclusion:

LLMs are not generative agents of meaning. They are simulators of pattern. Their outputs instruct us about the limits of probabilistic description and the enduring necessity of first-order, answerable acts. Observing them, properly framed, reinforces the grain of instantiation and the relational cut — the heart of the ontology.

The Grain of Instantiation IV (A Faculty Dialogue)

Setting: Faculty common room, late morning. Dr Finch has joined the discussion, notebook in hand. Professor Quillibrace, Mr Blottisham, and Miss Elowen Stray are present.

---

Finch:

If I may — two questions arise. First, Halliday often speaks of potential as probabilistically weighted and updated with each instantiation. How does that sit with the idea that probability survives the act rather than drives it?

Quillibrace:

It sits quite comfortably. Probabilities model the distribution of past acts — second-order patterning. They do not generate or constrain the first-order act itself.

Blottisham:

But surely they must bias or guide subsequent choices?

Stray:

Not really. They condition the space of possibilities, but they do not select the act. Answerability remains first-order.

---

Finch:

Second, Halliday says language construes context as well as realising it. Does that imply context participates directly in meaning?

Stray:

Only as it is experienced. Context-as-phenomenon is always construed in the act. Context-as-organisation conditions, but does not determine. They remain ontologically distinct.

Quillibrace:

Exactly. Language realises potential while simultaneously apprehending situational experience. The two are related, but the act of meaning itself is always first-order and irreducible.

Finch:

So probabilistic potential and context are always one step removed from meaning, shaping the arena but not producing the act?

Quillibrace:

Precisely. Probability explains traces, context explains recognisability, but neither produces meaning. That is always the purview of the act.

Blottisham:

And here I thought I had captured it fully.

Stray:

You captured patterns. But the act itself remains answerable — and that is what this ontology refuses to let collapse.

(The discussion continues in quiet contemplation, the notebook full but the key distinction firmly in place.)

The Grain of Instantiation III (A Faculty Dialogue)

What Holds

Setting: The faculty common room. Late afternoon. The blackboard bears a single line, freshly written:

ACT ≠ PATTERN

Professor Quillibrace sits with a book closed on his lap.

Mr Blottisham stands near the board, arms folded.

Miss Elowen Stray sits on the table, watching both.

---

Blottisham:

All right. I’ll concede most of it. Probability doesn’t cause meaning. Context doesn’t force it. Fine.

Quillibrace:

Good.

Blottisham:

But surely they still explain it — in aggregate.

Stray:

Explain what exactly?

Blottisham:

Why this meaning occurred rather than another.

Stray:

No.

(A pause.)

Blottisham:

No?

Stray:

They explain why this meaning was possible, recognisable, survivable. They don’t explain why it was done.

Quillibrace:

That distinction matters.

Blottisham:

But if the space is sufficiently narrowed—

Stray:

—something still has to happen in the space.

(She gestures to the board.)

Stray:

Patterns describe where acts tend to fall. They don’t perform the falling.

Blottisham:

So meaning isn’t selected?

Stray:

Not unless you’ve already smuggled in a selector.

Quillibrace:

Selection is a second-order metaphor. Meaning is first-order action.

Blottisham:

Then what exactly disappears when we reduce meaning to probability?

Stray:

The moment of answerability.

Blottisham:

Responsibility again.

Stray:

Not “again”. Always.

(Silence.)

Quillibrace:

Meaning is not what tends to occur. It is what is done.

(Blottisham nods, slowly.)

---

What Slips

Setting: A corridor outside a seminar room. The next day.

Mr Blottisham speaks animatedly to Dr Finch, a junior colleague clutching a notebook.

---

Blottisham:

It’s quite elegant, really. You see, probability doesn’t determine meaning — that’s the key insight.

Finch:

So meaning isn’t probabilistic?

Blottisham:

Oh, it is — but not in a crude way. It emerges from constrained choice.

Finch:

Choice by whom?

Blottisham:

Well — by the system-user, of course, operating within statistically structured context.

Finch:

So the probabilities guide the act?

Blottisham:

Guide, bias, channel — yes, exactly.

Finch:

And that explains meaning?

Blottisham:

Explains why certain meanings occur rather than others.

(They pause near a window.)

Finch:

Then if we had the probabilities and the context fully specified—

Blottisham:

—we’d have a very strong account of meaning, yes.

(Miss Stray passes by, overhearing the last sentence. She stops.)

Stray:

No, you wouldn’t.

Blottisham:

Elowen — I was just explaining—

Stray:

I know. You were explaining it correctly, and saying it wrong.

Finch:

I’m lost.

Stray:

You’ve turned conditions into inclinations.

Blottisham:

That’s a bit unfair.

Stray:

Is it? You’ve let probability lean again.

Blottisham:

But I explicitly said it doesn’t determine!

Stray:

You let it interact.

(A pause.)

Stray:

Probability doesn’t guide acts. It survives them.

Finch:

So what does the act respond to?

Stray:

To a situation — as construed — and to others who can answer back.

Blottisham:

Then what are probabilities for?

Stray:

For us.

(She gestures back toward the seminar rooms.)

Stray:

They are how we talk about what tends to happen after meaning has already occurred.

(She leaves.)

---

The mistake persists not because it is crude, but because second-order descriptions are so easy to mistake for agents.

The Grain of Instantiation II (A Faculty Dialogue)

 Setting:

Same faculty room. Evening now. The light is dimmer. The teapot has been replaced by glasses. The blackboard has been wiped clean, except for a single phrase written in chalk:

ACT ≠ PATTERN

Quillibrace sits, relaxed.

Stray is perched on the edge of the table.

Blottisham stands closer to the board than before.

---

The Almost

Blottisham:

All right. I think I see it now. The system doesn’t cause the act. The probabilities don’t force the meaning.

Quillibrace:

Good.

Blottisham:

They just… make certain selections more likely.

Quillibrace:

(Still calm)

Careful.

Blottisham:

Likely in the sense that they’re already there. Already sedimented. So when someone speaks, they’re drawing on what’s most available.

Stray:

Available to whom?

Blottisham:

To the system.

Quillibrace:

There it is.

(A pause.)

---

The Slide

Blottisham:

No, wait—hear me out. I’m not saying the system acts. I’m saying the act is… statistically anticipated.

Quillibrace:

Anticipated by whom?

Blottisham:

By the distribution itself.

Stray:

So the distribution expects something?

Blottisham:

Not expects—biases.

Quillibrace:

Bias is already too strong.

Blottisham:

Surely not. Bias just means unevenness.

Quillibrace:

Unevenness in description, yes. Not inclination in the world.

Stray:

You’re letting the map lean.

(Blottisham frowns, rubs his forehead.)

---

The Almost-Insight

Blottisham:

So the act… the act is free?

Quillibrace:

It is answerable.

Blottisham:

But not unconstrained?

Quillibrace:

Of course not.

Blottisham:

Then the system guides the act.

Quillibrace:

It conditions the space in which the act can occur.

Blottisham:

That sounds like guidance.

Stray:

It sounds like weather.

(A small smile from Quillibrace.)

---

Where He Fails

Blottisham:

So meaning emerges from the interaction between agency and probability.

Quillibrace:

No.

Blottisham:

Why not? That seems fair.

Quillibrace:

Because probability does not interact.

Blottisham:

Surely it does—look at learning, adaptation, updating—

Quillibrace:

Those are properties of descriptions, not of acts.

Stray:

You’ve moved the interaction to a place where nothing can answer back.

(Blottisham opens his mouth, then closes it.)

---

The Diagnostic Moment

Blottisham:

So close… I can almost say it.

Quillibrace:

Say it anyway.

Blottisham:

Meaning is… selected under constraints?

Quillibrace:

And now you’ve lost it.

Blottisham:

Because?

Stray:

Because selection already assumes a chooser.

(Silence.)

---

The Quiet End

Blottisham:

So probability never touches meaning.

Quillibrace:

It touches its traces.

Stray:

And confuses us because the traces are all that remain.

(Blottisham looks at the chalk on the board.)

Blottisham:

ACT ≠ PATTERN.

Quillibrace:

You can understand that sentence.

Stray:

But you still want the arrow.

(Blottisham nods, rueful.)

---

Curtain.

The Grain of Instantiation I (A Faculty Dialogue)

 Setting:

The senior common room. Late afternoon. A tray with a teapot, three cups. The blackboard still bears half-erased diagrams of systems and instances. Light filters in, dust motes visible.

Professor Quillibrace sits, reading glasses low on his nose.

Mr Blottisham stands, jacket off, coffee in hand.

Miss Elowen Stray leans against the window, notebook closed but present.

---

The Grain of Instantiation

Blottisham:

Surely—surely—if we can model the probabilities finely enough, we’ve captured what’s really going on. I mean, the grain, the tendencies, the regularities—what more is there?

Quillibrace:

(Without looking up)

An act.

Blottisham:

An act?

Quillibrace:

Yes. One. Occurring. Irreducible.

Blottisham:

But that’s just mystification. The act comes from the system. The probabilities are updated with every instance—Halliday himself says so.

Quillibrace:

Halliday says the description is updated. Not the ontology.

Blottisham:

That’s a distinction without a difference.

Stray:

(Quietly)

Is it? Or is it the difference between what remains and what happens?

(Blottisham pauses, frowns.)

---

First- and Second-Order

Blottisham:

Look—corpora show us what language is. The patterns are there. Massive scale. Surely that’s meaning made visible.

Quillibrace:

Meaning made absent, Mr Blottisham. What you see are residues.

Blottisham:

Residues of what?

Quillibrace:

Of acts that are no longer there.

Stray:

And the acts themselves—once they’re gone—they can’t be recovered from the pattern, can they?

Quillibrace:

No. Only inferred. And inference is not instantiation.

Blottisham:

But the pattern constrains what can be said next.

Quillibrace:

It conditions. It does not determine.

Stray:

That word keeps doing important work.

---

Context and Conditioning

Blottisham:

All right, context then. Field, tenor, mode. Surely once you specify those, the meaning more or less follows.

Quillibrace:

If that were true, responsibility would evaporate.

Blottisham:

Oh come on—context explains meaning.

Quillibrace:

It explains why certain meanings are recognisable. Not why this one occurred.

Stray:

So context narrows the space… but doesn’t select the point?

Quillibrace:

Exactly.

Blottisham:

But Halliday says language construes context!

Quillibrace:

And so it does.

(Quillibrace finally looks up.)

Quillibrace:

Context as experienced is construed. Context as organisation is encountered.

Blottisham:

That sounds convenient.

Stray:

Or careful.

---

Agency and Acts

Blottisham:

You’re smuggling in agency. That’s the real issue. You don’t want to give it up.

Quillibrace:

I’m not smuggling it in. I’m refusing to remove it.

Blottisham:

Systems act. Institutions act. Models act.

Quillibrace:

No. They enable, constrain, stabilise. They do not answer.

Stray:

And answering is what makes something an act?

Quillibrace:

Yes. Answerability is not optional. It is constitutive.

Blottisham:

So a language model can never mean?

Quillibrace:

It can only be meant with.

(A brief silence.)

---

The Grain Revisited

Stray:

May I try something?

Quillibrace:

Please.

Stray:

It seems the grain of instantiation is where everything gets confused. The finer the grain we observe, the more we’re tempted to think we’re seeing the act itself.

Quillibrace:

Well put.

Stray:

But what we’re really seeing is the trace density of past acts.

Blottisham:

And that isn’t enough?

Stray:

Not if what you’re trying to explain is meaning.

(Blottisham exhales, half-laughs.)

---

What Disappears

Blottisham:

So what, then, disappears when we reduce everything to probability?

Stray:

(After a pause)

The moment where something could have been otherwise—and wasn’t.

Quillibrace:

And with it, responsibility.

(Quillibrace closes his book.)

Quillibrace:

Probability explains fluency.

Context explains recognisability.

Institutions explain durability.

(He stands.)

Quillibrace:

None of them explains meaning.

---

Curtain.

Relational Ontology: Two Hallidayan Questions Clarified

This short post addresses two questions that an SFL scholar might naturally raise in response to the consolidated statement of relational ontology. The purpose here is not to revise the ontology, but to clarify how core Hallidayan insights are preserved and sharpened once the relevant ontological cuts are made explicit.

---

1. If potential is probabilistically weighted and updated with each instantiation, does probability not belong to meaning itself?

Halliday’s claim that linguistic potential is probabilistically weighted is both empirically grounded and theoretically powerful. It captures the fact that systems of meaning exhibit patterned tendencies, and that these tendencies shift as language is used.

The crucial clarification is ontological.

Probabilistic weighting belongs to the description of semiotic potential, not to the production of meaning. Probabilities model the distribution of past instantiations; they do not generate or drive new ones.

Each act of meaning leaves a residue that contributes to second-order patterning. Over time, this alters the statistical texture by which the system is described. In this sense, potential can be said to be “updated” with each instantiation.

What does not follow is that probability exerts causal force over meaning. Instantiation remains a perspectival cut from potential to event. Probability describes the grain of that cut after the fact; it does not supply the principle of actualisation.

Read this way, Halliday’s probabilistic system is entirely compatible with a relational ontology: probability belongs to modelling, not to meaning.

---

2. If language construes context as well as realising it, does this not collapse context into semantics?

Halliday’s claim that language construes context has often been misunderstood as a form of semantic idealism. In fact, it is best read as a claim about experience rather than determination.

Context exists in two analytically distinct but related ways.

First, context exists as phenomenon: situations as they are experienced, recognised, and made meaningful. In this sense, context is necessarily construed. There is no unconstrued situation, just as there is no unconstrued meaning.

Second, context exists as organised potential: historically sedimented social organisation, institutional structures, and patterned expectations. In this sense, context is not construed in the act. It conditions the act by delimiting what is recognisable, appropriate, or likely.

Language therefore both:

• realises contextual potential (field, tenor, mode), and

• construes contextual experience in first-order acts of meaning.

What it does not do is generate context as a determining system or collapse contextual structure into semantics. Conditioning and construal remain distinct.

---

3. Why These Clarifications Matter

These two questions mark the principal points at which probabilistic and contextual accounts are tempted to overreach.

By enforcing the distinctions between:

• first- and second-order phenomena,

• description and production,

• conditioning and determination,

the relational ontology preserves Halliday’s insights while removing the ambiguity that allows reductionist readings to arise.

The result is not a departure from SFL, but a clarification of its ontological commitments: meaning remains enacted, situated, and answerable, while probability and context retain their proper explanatory roles.

Nothing essential is lost. Much confusion is avoided.

Relational Ontology: A Consolidated Statement

This post consolidates the relational ontology as it now stands. It does not introduce a new trajectory or advance a polemic. Its purpose is constitutive: to restate the ontology formally, incorporating recent refinements and making its internal asymmetries explicit.

The ontology is relational throughout. Nothing here presupposes substances, representations, or meanings that exist independently of construal.

---

1. System and Instance

A system is structured potential: a theory of possible instances. It is not a container, a mechanism, or an agent. It is a space of organised possibilities.

An instance is not produced by the system. It is the system taken up from a perspective. Instantiation is therefore not a temporal process or a causal transition, but a perspectival cut from potential to event.

System and instance are not two kinds of thing. They are two relational moments of the same semiotic organisation.

---

2. Instantiation and Grain

Instantiation always occurs at a particular grain. Acts of meaning resolve possibilities at varying degrees of fineness relative to historically sedimented patterns of use.

This grain is not constitutive of meaning itself. It is a property of how instantiations distribute over time and across populations.

Probabilistic descriptions model this grain. They describe the density and regularity of past instantiations. They do not generate, explain, or determine meaning.

Probability therefore belongs to the afterlife of instantiation, not to its source.

---

3. Construal and Phenomenon

There is no unconstrued phenomenon. Construal is constitutive of experience and of meaning.

A phenomenon is first-order construed experience. It exists only as enacted and recognised in an act of meaning.

There is no meaning prior to construal, and no construal without instantiation.

---

4. First- and Second-Order Asymmetry

The ontology distinguishes sharply between:

• first-order phenomena: acts of meaning as construed experience, and

• second-order patterning: abstractions over multiple acts (e.g. corpora, probabilities, norms, distributions).

This distinction is ontologically asymmetric.

Second-order patterning is dependent on first-order acts. It cannot feed back into meaning without a new act of construal. Patterns do not explain meaning; they presuppose it.

Attempts to reduce meaning to patterning are category errors arising from the visibility of second-order residues.

---

5. Context as Conditioning

Context is organised semiotic potential. In a Hallidayan sense, field, tenor, and mode are contextual variables realised by semantics.

Context conditions acts of meaning. It constrains what is likely, conventional, or recognisable. It does not determine what is meant.

No accumulation of contextual variables converts conditioning into determination. Context enriches probability; it does not replace agency or construal.

---

6. Meaning as Act

Meaning is not an outcome, a state, or a system property. Meaning occurs in acts.

An act of meaning involves:

• construal,

• uptake,

• and answerability.

Agency and responsibility are therefore not external ethical overlays. They are internal to meaning itself. Without the possibility of answerability, there is no meaning.

Systems, institutions, and technologies do not act. Only persons act.

---

7. Institutions and Coordination

Institutions are not agents. They are organised, historically sedimented semiotic potential.

They stabilise expectations, distribute roles, and coordinate acts across time and space. They enable scale by multiplying the conditions under which acts can occur.

Institutions do not produce meaning. They condition, channel, and constrain the instantiation of meaning by persons.

Scale does not collapse acts into systems. It aggregates instantiations without eliminating their first-order status.

---

8. Technology and Probability

Technologies that operate over language — including statistical and machine-learning systems — operate exclusively at the level of second-order patterning.

They model distributions of past acts and reproduce their grain with great efficiency. They do not construe, act, or mean.

Their outputs acquire meaning only when taken up in new acts of construal by persons.

---

9. Boundary Conditions and Stability

The ontology maintains its stability by enforcing a set of non-negotiable asymmetries:

• system / instance,

• potential / event,

• first-order / second-order,

• conditioning / determination,

• coordination / agency.

These distinctions are relational, not ontic partitions. They explain both the power of probabilistic and systemic descriptions and the limits beyond which they fail.

The ontology therefore predicts its own misreadings and explains its resistance to collapse.

---

10. Closing

Meaning happens. It is enacted, situated, answerable, and irreducibly first-order.

Everything else — probability, context, institutions, coordination, technology — describes, conditions, or stabilises the conditions of its occurrence.

Nothing replaces the act.

This is the relational ontology as it now stands.

The Grain of Instantiation: Series Summary

After six posts tracing the trajectory from fluency to meta-theoretical reflection, it is useful to pause and map the architecture of the argument. This summary consolidates the cuts, asymmetries, and relational logic that make the ontology robust.

---

1. Fluency and Grain

• Observation: Probabilistic patterns (e.g., corpora, statistics, LLM outputs) reveal fluency and repetition.

• Cut: Fluency explains the grain of instantiation, not the source of meaning. Probability describes residues of acts, not the acts themselves.

2. Phenomena vs Patterning

• Observation: Language leaves traces, but these traces are second-order.

• Cut: First-order phenomena (construals) are irreducible; second-order patterning (probabilities, corpora) presupposes them. Patterning cannot generate meaning.

3. Context and Conditioning

• Observation: Situations, fields, tenors, and modes shape what is possible.

• Cut: Context conditions construal without determining it. Situational enrichment refines probability but does not create acts of meaning.

4. Meaning as Act, Not Outcome

• Observation: Meaning emerges in use, with intent, recognition, and answerability.

• Cut: Meaning is an act. Agency and responsibility are internal to meaning. Systems and technologies cannot be agents.

5. Coordination and Scale

• Observation: Institutions, genres, and coordination stabilize acts across time and space.

• Cut: Coordination enables scale without replacing the act. Institutions organise potential but do not enact meaning. Scale multiplies acts without collapsing them into systemic outcomes.

6. Meta-Theoretical Apex

• Observation: The ontology resists reduction because it explicitly encodes asymmetries and dependencies.

• Cut: System–instance, first–second order, act–outcome, conditioning–determination, coordination–agency. These cuts are relational and structural. They preserve meaning against collapse.

---

The Boundary-Maintaining Principle

Meaning always occurs in acts. Probability, context, coordination, scale, and technology describe, condition, or scaffold these acts—but they never replace them. This is the central insight that unites the series and renders the ontology unassailable.

---

Why This Matters

• Analytical clarity: We can discuss LLMs, institutions, and fluency without confusing residue for meaning.

• Normative clarity: Responsibility remains grounded in acts, not systems.

• Pedagogical clarity: The architecture can be diagrammed, taught, and applied.

The Grain of Instantiation is now visible in full: the cuts, the asymmetries, and the relations. Meaning happens. It is enacted. It is answerable. And the series shows exactly why it cannot be reduced, predicted, or outsourced to systems.