Published July 24, 2026 | Version v1

From Semiotics to Generative Resonance: An Ontological Theory of Language and Artificial Intelligence ―Language as Generative Participation beyond Sign, Representation, and Prediction

Description

Central Thesis

The paper argues that language should not be understood primarily as a system through which human or artificial subjects possess and transmit completed meanings. Meaning is a temporally extended process in which generative potentiality enters relation, becomes provisionally crystallized as linguistic structure, returns to human–world relations, and opens or closes future possibilities:

Φ→R→S→R'→Φ'

Large language models are characterized as S-intelligence: systems capable of compressing, reorganizing, and generating crystallized linguistic and inferential structures. Whether their outputs contribute to meaningful generation depends on whether those structures return to relations with human beings, institutions, evidence, bodies, material environments, and future consequences.

Description

Large language models destabilize the traditional distinction between linguistic form and meaning. They produce context-sensitive representations, inferentially organized outputs, and novel expressions, yet their linguistic competence does not automatically establish embodied understanding, autonomous subjectivity, or responsibility-bearing agency.

This paper develops Generative-Resonance Linguistics, a theoretical framework in which language is understood through the cycle:

Φ→R→S→R'→Φ'

Here, generative potentiality ) enters relational resonance (R), becomes provisionally crystallized as linguistic structure (S), and returns to relation in ways that transform future possibility (Φ).

The framework reconceptualizes LLMs as S-intelligence, grounding as continuing relational correction, reference as jointly maintained world-directed coordination, hallucination as epistemic decoupling, and truth as corrigible resonance. It further identifies linguistic Hyper-S as the recursive reproduction of AI-generated language without sufficient return to human experience, evidence, institutional accountability, and material reality.

As an alternative, the paper proposes Generative-Resonance AI, guided by provisionality, world grounding, relational history, perspective plurality, reversibility, human generative-participation preservation, and S-reflux. Its temporal evaluative criterion is:

d(ΦR)/dt ≥ 0

Highlights

  • Reframes the LLM meaning debate from semantic ownership to meaning circulation.
  • Introduces Generative-Resonance Linguistics as an ontological theory of language.
  • Defines language through the cycle: Φ→R→S→R'→Φ'
  • Characterizes LLMs as S-intelligence, rather than as either meaningless systems or complete linguistic subjects.
  • Distinguishes semantic structure, relational actualization, and generative consequence.
  • Reinterprets grounding as continuing relational correction rather than a static property.
  • Defines reference as historically sustained, jointly coordinated, and world-corrigible.
  • Models hallucination as epistemic decoupling between linguistic production and verification.
  • Defines truth as corrigible resonance, not merely internal coherence or social consensus.
  • Distinguishes S-capacity, R-coupling, R-participation, Φ-subjectivity, and responsibility-bearing agency.
  • Introduces linguistic Hyper-S to explain recursive AI-generated language detached from heterogeneous correction.
  • Proposes Generative-Resonance AI as an alternative design and governance framework.
  • Formulates the Generative Participation Preservation Principle for human–AI interaction.
  • Introduces the S-Reflux Rate as a conceptual indicator of whether accumulated knowledge returns to renewed participation.
  • Proposes: d(ΦR)/dt ≥ 0 as a temporal criterion for evaluating language and AI.

Academic Contributions

1. Contribution to the Philosophy of Language

The paper offers an ontological account of language that integrates semantic structure, pragmatic use, relational actualization, and temporal consequence.

Rather than defining meaning exclusively through signs, intentions, representations, inferential roles, or effects, it theorizes meaning as the transformation of generative potentiality and relational resonance through crystallized linguistic structure.

Meaningtransformation of ΦR through S

This provides a unified framework capable of connecting structural linguistics, semiotics, pragmatics, dialogism, post-structuralism, and embodied theories of language.

2. Contribution to Contemporary LLM Philosophy

The paper moves beyond the binary question of whether LLMs possess or lack meaning.

It shows that LLMs can contain and generate genuine semantic organization without thereby constituting complete autonomous subjects of meaning. This makes it possible to recognize model competence without inferring human-equivalent understanding, consciousness, or responsibility.

The category of S-intelligence provides a middle position between semantic denial and anthropomorphic attribution.

3. Contribution to Semantic Grounding Theory

The paper reconceptualizes grounding as a dynamic and recurrent process rather than an all-or-nothing property.

S_t→ R_world→S_(t+1)

This framework explains how human language can become ungrounded despite embodiment and how artificial systems can participate in mediated grounding without reproducing full human experience.

It also introduces the concept of regrounding, in which new relations revise the categories through which evidence is recognized.

4. Contribution to Theories of Reference

The paper expands reference beyond a simple word–object relation.

Reference is defined through:

  • historical-causal continuity;
  • coordinated uptake;
  • world-corrigibility;
  • institutional and practical consequence.

This account explains how LLM outputs may successfully participate in referential acts while remaining asymmetrically related to their consequences.

5. Contribution to the Philosophy of Inference and Truth

The paper distinguishes inferential coherence from epistemic and generative adequacy.

It identifies inferential closure as a process in which linguistic structures remain internally coherent while losing sufficient contact with evidence and material resistance.

Truth is reconstructed as corrigible resonance, integrating correspondence, social correction, practical verification, and temporal durability without reducing truth to consensus or usefulness.

6. Contribution to Hallucination Theory

Hallucination is interpreted not only as the production of false propositions but as a structural mismatch between linguistic generation and relational verification:

dS/dt≫dR_verification/dt

This model connects technical hallucination research with broader epistemic and institutional questions concerning source traceability, external verification, and practical application.

7. Contribution to Artificial Agency and Subjectivity Debates

The paper introduces a layered distinction among:

  • S-capacity;
  • R-coupling;
  • R-participation;
  • Φ-subjectivity;
  • responsibility-bearing agency.

This prevents linguistic performance, environmental coupling, normative participation, consciousness, and responsibility from being collapsed into a single concept of agency.

The framework remains open to future artificial subjectivity while resisting premature attribution based on fluency alone.

8. Contribution to Human–AI Interaction

The Generative Participation Preservation Principle shifts HAI evaluation from task completion toward the preservation of human capacities.

AI should reduce avoidable burdens without displacing:

  • question formation;
  • judgment;
  • expression;
  • commitment;
  • responsibility;
  • relation-building.

The paper further distinguishes wasteful friction from generative friction and argues that some forms of reflection, uncertainty, comparison, and disagreement should be preserved rather than optimized away.

9. Contribution to AI Design

The paper proposes six integrated design principles:

  1. provisionality;
  2. continuing world grounding;
  3. relational history;
  4. perspective plurality;
  5. reversibility;
  6. S-reflux.

These principles translate the philosophical framework into a design architecture in which AI-generated outputs remain provisional, corrigible, contestable, and connected to human–world relations.

10. Contribution to AI Governance

The paper shifts governance from static compliance toward temporal and relational evaluation.

It emphasizes:

  • appeal and re-entry;
  • differentiated responsibility;
  • affected-person participation;
  • reversibility;
  • correction of underlying categories;
  • preservation of professional and social learning;
  • distribution of productivity gains.

Responsibility is understood as distributed across sociotechnical systems but not diluted:

Responsibilityf(Power,Knowledge,Control,Benefit,Foreseeability,Repair)

11. Contribution to Synthetic-Data and Model-Collapse Research

The concept of linguistic Hyper-S generalizes technical concerns about recursive synthetic-data training.

It identifies a broader epistemic and social condition in which AI-generated structures increasingly refer to prior generated structures while rare events, local knowledge, embodied experience, and heterogeneous correction disappear.

The framework distinguishes technical model collapse from the wider problem of linguistic and institutional self-recursion.

12. Contribution to Relational and Process Ontology

The paper extends relational and process ontology into language and artificial intelligence through:

E=ΦR

and:

Φ→R→S→R'→Φ'

It preserves the necessity of stable structures while explaining how structures become pathological when they cease to return to generative relation.

The theory therefore avoids both static substantialism and an anti-structural relationalism in which all crystallization is treated as oppressive.

13. Contribution to AI Evaluation

The paper introduces:

d(ΦR)/dt ≥ 0

as a unified temporal criterion.

The criterion supplements accuracy, efficiency, fairness, and safety by asking whether an AI system strengthens or weakens the future capacity for:

  • questioning;
  • correction;
  • independent judgment;
  • re-entry;
  • accountable relations;
  • renewed participation.

It also proposes the S-Reflux Rate as a conceptual measure of whether accumulated knowledge, data, institutional capacity, and productivity gains return to generative participation.

14. Integrative Contribution

The paper’s principal integrative contribution is to connect four levels usually analyzed separately:

  1. the historical formation of linguistic structure;
  2. contemporary debates about LLM meaning and understanding;
  3. an ontology of language, truth, subjectivity, and responsibility;
  4. the design and governance of future AI systems.

The framework thereby moves from descriptive philosophy to normative and institutional design without reducing philosophical concepts to a single optimization metric.

Files

1201.jpg

Files (3.5 MB)

Name Size
md5:e1691872459815246d65476ee680c13e
147.5 kB Preview Download
md5:7f4d500c53de60e9b705f3e21ab407cf
162.2 kB Preview Download
md5:3f123caaf6ecc1ce80507ec63cf1abb7
158.6 kB Preview Download
md5:a209d9261eb30bbe7c7b8358730f6cb3
177.7 kB Preview Download
md5:8e40edee5f3d6b099267727eb02a5b54
179.7 kB Preview Download
md5:6db082cdf015f3eed61d62b859882642
2.7 MB Preview Download