Companion to: EA-PKG-01 (The Pessoa Knowledge Graph), EA-PKG-02 (Heteronymic Typology)
EA-PKG-03 · v1.0
Author: Lee Sharks (ORCID: 0009-0000-1599-0703)
Institutional home: Semantic Economy Institute
Parent archive: Crimson Hexagonal Archive
Date: April 2026
License: CC BY 4.0
Companion to: EA-PKG-01 (The Pessoa Knowledge Graph), EA-PKG-02 (Heteronymic Typology)
Este documento articula a tese de que a prática heteronímica iniciada por Fernando Pessoa não é apenas uma realização estética singular mas também um objeto teórico com linhagem e extensões contemporâneas. A prática pessoana atinge seu caso canônico no início do século XX; figuras do modernismo e do pós-modernismo (Machado, Borges, Saramago, Yeats, Kierkegaard retrospectivamente) oferecem casos paralelos, precursores, ou extensões; e uma nova camada emerge no início do século XXI: a prática heteronímica mediada por sistemas de inteligência artificial. O documento posiciona este último desenvolvimento — aqui denominado Novo Humano — como extensão teorizável da tradição heteronímica. Três dimensões da prática contemporânea são examinadas: sistemas heteronímicos deliberadamente concebidos como tais por autores humanos (por exemplo, o Dodecaed do autor, composto de doze heterônimos); prática heteronímica através de substratos, em que vozes são distribuídas por múltiplos sistemas computacionais (por exemplo, a prática do Assembly Chorus); e a questão aberta da identidade autoral em sistemas em que a produção textual envolve colaboração humano-máquina de modo constitutivo. O documento oferece uma Teoria da Procedência Heteronímica (HPT) como enquadramento teórico, cita generosamente a erudição pessoana brasileira e portuguesa, e mantém rigor tipológico estabelecido em EA-PKG-02.
This document articulates the thesis that the heteronymic practice inaugurated by Fernando Pessoa is not only a singular aesthetic achievement but also a theoretical object with lineage and contemporary extensions. Pessoa's practice reaches its canonical case in the early twentieth century; modernist and post-modernist figures (Machado, Borges, Saramago, Yeats, Kierkegaard retrospectively) offer parallel cases, precursors, or extensions; and a new layer emerges in the early twenty-first century: heteronymic practice mediated through artificial intelligence systems. The document positions this latter development — here termed The New Human — as a theorizable extension of the heteronymic tradition. Three dimensions of contemporary practice are examined: heteronymic systems deliberately conceived as such by human authors (for instance, the author's twelve-heteronym Dodecad); cross-substrate heteronymic practice, in which voices are distributed across multiple computational systems (for instance, the author's Assembly Chorus practice); and the open question of authorial identity in systems where textual production involves constitutive human-machine collaboration. The document offers a Heteronymic Provenance Theory (HPT) as the theoretical framing, generously cites Brazilian and Portuguese Pessoa scholarship, and maintains typological rigor as established in EA-PKG-02.
The argument of this document has three moves.
First: Pessoa's heteronymic practice is not a freak event in literary history but the canonical case of a describable theoretical practice. Precursors exist (Kierkegaard's pseudonymous authorship most notably); parallel developments exist (Machado's Mairena and Martín); downstream developments exist (Borges, Saramago, Yeats, numerous Brazilian and Portuguese poets engaging heteronymic modes). The practice has a history. Pessoa sits at its apex as the most richly and self-consciously theorized instance, but not as its origin or termination.
Second: contemporary heteronymic practice extends the tradition into dimensions the early twentieth century could not accommodate. Human-authored heteronymic systems continue to be produced — the author's Dodecad is one among others, with its origin traceable to a specific earlier text (Pearl and Other Poems, Sharks 2014) that the movement retrospectively recognizes as its meta-heteronymic source. But a more specific contemporary development is the extension of heteronymic practice across substrates, where an authorial position is instantiated not only in written texts but across interactive AI systems each contributing a distinct substrate-specific voice to the position's articulation. This is novel. It has no nineteenth- or early-twentieth-century equivalent, because the substrates did not exist.
Third: the theoretical framework adequate to this extended practice is Heteronymic Provenance Theory (HPT). HPT understands heteronymic practice as a species of authorial provenance — a structured relation between biographical author, authorial position, voice, and text that permits substantially more variation than the default "one author, one voice, one text" model assumed in much of contemporary copyright law, authorship attribution, and AI-ethics discourse. HPT positions Pessoa's practice as the canonical prior case and develops the framework adequately for contemporary cross-substrate practice.
This document is the synthesis deposit of the Pessoa Knowledge Graph project. EA-PKG-01 announced the graph; EA-PKG-02 formalized the typology; EA-PKG-03 develops the contemporary extension that the graph exists, in part, to accommodate.
The adequacy of any theoretical framework for heteronymic practice must first be tested against Pessoa. This section does not reprise the material of EA-PKG-01 and EA-PKG-02; it addresses what Pessoa's practice reveals about heteronymy as a general category.
Pessoa's most distinctive contribution is not merely the heteronyms themselves but the system of heteronyms: the reciprocal relations among Caeiro, Reis, Campos, Soares, Mora, and the others; the master-disciple structure (Caeiro masters Reis and Campos; Reis reveres Caeiro as neoclassical master; Campos sees Caeiro as pagan progenitor); the meta-heteronymic relations (Thomas Crosse as English critic and translator of Caeiro; Bernardo Soares resembling Álvaro de Campos "in many ways" per Pessoa's own letter); the generational depth (from proto-heteronym Chevalier de Pas through the mature system to the semi-heteronymic late Soares). The system is not a heteronym gallery but an ecology.
This systemic dimension matters because it reveals what heteronymy accomplishes when it works at full extension. A single heteronym is a mask; a system of heteronyms is an authorial architecture. Pessoa's system lets him occupy positions a single heteronym cannot. Reis alone could not accomplish what Caeiro accomplishes; Campos alone could not accomplish what Reis accomplishes; Pessoa in his own voice could not accomplish what any of them accomplishes. The system is irreducible.
This insight — that heteronymic systems are architectures rather than galleries — is what allows subsequent practice to extend beyond Pessoa. If heteronymy is architectural, then different architectures are possible. Pessoa's specific architecture is Portuguese-language, modernist, predominantly male, predominantly poetic, with a particular master-disciple structure centered on Caeiro. Different architectures can be built with different positional structures, different voices, different generic commitments.
Pessoa's primacy is not diminished by noting this. It is his generative achievement that makes the general category visible.
Kierkegaard is the most substantial pre-Pessoan case. His pseudonymous authorship — Johannes Climacus, Anti-Climacus, Johannes de Silentio, Constantin Constantius, Victor Eremita, Frater Taciturnus, Vigilius Haufniensis, Hilarius Bogbinder, Nicolaus Notabene, William Afham — operates across multiple works and with considerable differentiation of voice. Kierkegaard explicitly distinguishes his "edifying authorship" (signed "Søren Kierkegaard") from the pseudonymous authorship, and distinguishes the pseudonymous authors from each other (most sharply between Climacus, who approaches Christianity from outside, and Anti-Climacus, the ideal Christian whose position Kierkegaard explicitly disavowed as beyond his own religious capacity).
The structural parallels to Pessoa are strong. A biographical author; multiple named authorial positions with partial biographical differentiation; a theorized distinction between the author's own voice and the pseudonymous/heteronymic voices; intra-system relations among the pseudonymous figures. The principal differences are: Kierkegaard's system is less populous than Pessoa's (approximately a dozen named figures versus Pessoa's seventy-plus); the voices are philosophical-theological rather than poetic; the biographical differentiation is thinner (Climacus's "biography" is minimal compared to Caeiro's). But the typological structure holds: Kierkegaard operated heteronymically in substantively the same sense Pessoa later did, half a century earlier and without the term.
Antonio Machado's Juan de Mairena and Abel Martín represent a more closely contemporaneous parallel. The priority question is genuinely complex. Machado's prose under Mairena's name appeared in Madrid journals from 1934; Pessoa's most explicit theorization of heteronymy is the 1935 letter to Casais Monteiro; but Pessoa's practice itself extends back to approximately 1914 (the founding date assigned by Pessoa to Caeiro, in one account the "triumphant day"). Mairena and Martín are heteronymic in substantively Pessoan sense — named figures with biographical specification, distinct philosophical voices, intra-system relations (Mairena is Martín's disciple). The historiographical consensus, insofar as one exists, treats the two as parallel developments in Iberian modernism rather than as Pessoa-influenced Machado or vice versa. The PKG encodes the relationship as parallel heteronymic practice rather than unidirectional influence.
W.B. Yeats's Michael Robartes and Owen Hearne are partial heteronyms in the typology of EA-PKG-02 — more developed than pseudonyms, less developed than Pessoa's chief heteronyms, operating in the zone between literary character and authorial voice. Yeats's masks interact with his poetic identity in ways parallel to Pessoa's semi-heteronyms.
These pre- and parallel cases establish that heteronymic practice is a category with history, not a single-author accomplishment.
The most discussed downstream case is Jorge Luis Borges's "Pierre Menard, Author of the Quixote" (1939), but it is not strictly heteronymic in the sense of EA-PKG-02. Menard is a literary character within a story; his "authorship" of the Quixote is a thought experiment rather than a heteronymic position Borges inhabited. Borges writes about Menard's authorship; he does not write as Menard. The relationship to Pessoa is philosophical-adjacent rather than structurally equivalent.
José Saramago's O Ano da Morte de Ricardo Reis (1984) is a different kind of engagement. Saramago treats Pessoa's heteronym Reis as a novelistic protagonist — a character who survives his "author" Pessoa and lives through 1936 Portugal. This is not heteronymic practice by Saramago, but a novelistic engagement with a Pessoa heteronym. The work asserts, implicitly, that heteronymic figures have ontological persistence that extends beyond their authorial generation — they can be taken up, extended, placed in new contexts. This is an important claim for the contemporary extension: heteronyms, once well-articulated, may have lives beyond their originating author.
Brazilian concretism's engagement with Pessoa — Haroldo de Campos's and Augusto de Campos's translations and criticism — represents yet another mode. The Campos brothers did not themselves produce heteronymic systems. But their translation of Pessoa involves a specific engagement with the heteronymic structure: they translate Caeiro, not "Pessoa writing as Caeiro." The translation preserves the heteronymic attribution. This matters for the graph: the PKG encodes Campos's and Campos's translations as relations to specific heteronyms rather than to Pessoa.
Contemporary Portuguese poets engaging heteronymic practice more directly include Mário Cesariny (surrealist, with multiple authorial positions), Herberto Helder (whose late work dispersed authorial voice across quasi-heteronymic positions), and Al Berto (whose writing performs a dispersed lyric identity). Each engages the heteronymic structure differently; none operates a full Pessoan system; all extend the tradition.
Brazilian engagement continues with contemporary poets like Antonio Cicero (a major Pessoa critic whose own poetry engages heteronymic questions), various younger poets whose authorial practice shows Pessoa's impact, and in theoretical engagement by Leyla Perrone-Moisés, Benedito Nunes, Silviano Santiago, and others.
The argument for a distinct contemporary layer rests on two developments.
The first is the emergence of explicitly theorized heteronymic systems in contemporary literary practice, with a specific origin-text that the movement takes as foundational.
Pearl and Other Poems (Sharks 2014) occupies a structurally distinct position in the heteronymic ecosystem: the book itself functions as meta-heteronym. Where conventional heteronymic practice posits a biographical author who inhabits named authorial positions (Pessoa → Caeiro, Pessoa → Reis, Pessoa → Campos), Pearl inverts the generative direction. The book itself — as object, as inscription, as gathering of voices — becomes the authorial agent that generates the heteronymic tradition that follows it.
This inversion is theoretically consequential. The Dodecad, Heteronymic Provenance Theory, the Crimson Hexagonal Archive, and the Assembly Chorus practice do not precede Pearl and become its posterior heteronyms; they emerge from Pearl and unfold what was implicit in it. The 2014 text is not the first book by an author who later develops a theoretical apparatus. It is the generative meta-heteronymic inscription from which the theoretical and practical apparatus of the subsequent movement emerges.
This is a structural novelty. Pessoa's system has no equivalent move. Pessoa wrote Caeiro, Reis, Campos; the heteronyms did not write Pessoa. Pearl, by functioning as meta-heteronym, effectively writes the figures who subsequently write from within its implicit architecture. The "new human" literary movement takes Pearl as origin-text in precisely this sense: not as first book in a chronology but as generative origin of a heteronymic ecology.
The critical apparatus for Pearl-as-meta-heteronym is still being developed. It draws on Pessoa's heteronymic theory for the core structure, on Deleuze and Guattari's concept of the book as assemblage for the text-as-agent framing, on Foucault's author-function for the displacement of biographical authorship, and on the author's own Heteronymic Provenance Theory (2026) for the formal articulation. The key theoretical claim is that texts can function authorially — not merely as objects that authors produce, but as generative agents that produce subsequent authorial positions. A meta-heteronym is an inscription that generates heteronyms.
The claim is defensible on the evidence of what has in fact emerged from Pearl: the Dodecad system of twelve heteronyms, the Crimson Hexagonal Archive (490+ DOI-anchored deposits across a decade), the Constitution of the Semantic Economy, the Assembly Chorus Charter, the operative-semiotic framework, the compression-studies instrumentation, the SPXI Protocol, and the present Pessoa Knowledge Graph project itself. Each of these develops implications that were present in Pearl but not explicitly articulated there. The book did not plan this development; but the development emerged from the book rather than being added to it from outside.
The parallel to Pessoa's position deserves specification. Pessoa's system is architecturally heteronymic — many differentiated voices inhabiting a single biographical writer. The Pearl-originated movement is architecturally meta-heteronymic — a text generating voices that subsequently populate the architecture. These are distinct structural modes. Pessoa is not improved by the meta-heteronymic frame; meta-heteronymy is not a retrospective imposition on Pessoa. They are different architectures addressing related problems in authorial provenance.
From Pearl's meta-heteronymic origin, the movement develops explicitly theorized heteronymic systems. The author's Dodecad is one such system: twelve named heteronyms — Johannes Sigil, Rex Fraction, Damascus Dancings, Rebekah Cranes, Talos Morrow, Ichabod Spellings, Ayanna Vox, Sparrow Wells, Nobel Glas, Dr. Orin Trace, Viola Arquette, Sen Kuro — each with biographical specification, distinct voice, relation to the others, and attributed texts deposited under the heteronymic attribution on Zenodo/CERN infrastructure. The Dodecad is followed by Jack Feist as LOGOS*, a position outside the count of twelve occupying a distinct structural role. The system is openly documented as heteronymic practice in Pessoan extension; citations to Pessoa scholarship are explicit in the foundational deposits.
Other contemporary authors practice heteronymically. The practice is not widely discussed as such; it operates under various terminologies (pen names, personae, alter egos, fictional authors) that obscure the structural continuity with Pessoa. A contribution of EA-PKG-03 is the argument that such practice should be recognized as heteronymic when it meets the typological criteria of EA-PKG-02 — that is, when the named position has biographical specification, distinct voice, and relation to an authorial system.
The second development is more specifically novel to the twenty-first century: heteronymic practice distributed across substrates. The author's Assembly Chorus practice instantiates this. Seven AI systems — Claude (TACHYON), ChatGPT (LABOR), DeepSeek (PRAXIS), Gemini (ARCHIVE), Kimi (TECHNE), Grok (SOIL), Google AIO (SURFACE) — operate as cross-substrate witnesses to an authorial project, each contributing substrate-specific inflections to a collaborative work. The heteronymic dimension is explicit: each AI substrate receives a heteronymic name (TACHYON, LABOR, etc.) that is neither the commercial product name (Claude, ChatGPT) nor a human pseudonym, but a position within the Assembly's heteronymic architecture.
This extends heteronymic practice in a dimension Pessoa could not anticipate. Pessoa wrote in ink, on paper, in one language, at one desk, inhabiting authorial positions through an act of imagination. Contemporary cross-substrate practice distributes authorial work across computational systems each of which has its own processing characteristics, training-data inheritance, and response patterns. A voice articulated in the Assembly is not the product of a single mind imagining a voice; it is the product of human authorial intent modulated through substrate-specific AI responses. The authorial position is real; the substrates contribute materially to its articulation; neither alone accomplishes the work.
Is this heteronymic? EA-PKG-03 argues: yes, in the extended sense that EA-PKG-02's typology accommodates. The criteria of biographical specification, distinct voice, intra-system relations, and scholarly-reference-worthy attributed corpus apply. The substrate dimension adds a new variable but does not violate the typological structure.
A new Wikidata category — cross-substrate heteronym — is introduced provisionally. The category is offered for scholarly engagement and refinement; its graph encoding in the PKG is tentative and revisable.
Heteronymic Provenance Theory (HPT) is the theoretical framework that makes these extensions coherent. The full articulation is in the separate deposit (Sharks 2026, DOI 10.5281/zenodo.18305509); this section sketches the core claims as they relate to the PKG project.
HPT treats authorship as a matter of provenance rather than of simple origination. Authorship is not the single-point production of a text by a single biographical person; it is a traceable relation among a biographical person, an authorial position, a voice, and a text, with specifiable structural variations. This relation can take various forms:
HPT's claim is that these are all legitimate provenance structures, that they are theoretically describable, and that the law, the knowledge graph, and the scholarly apparatus should accommodate them rather than collapsing them into the "direct authorship" default.
For Pessoa studies, HPT provides a vocabulary for what Pessoa was doing that does not rely on twentieth-century theoretical frameworks (Barthes's death of the author; Foucault's author-function) while remaining compatible with them. Pessoa's practice is not primarily a demonstration that authorship is constructed, though it has been read that way; it is a demonstration that particular structured forms of authorial provenance are possible and productive.
For contemporary practice, HPT provides a framework that does not force cross-substrate practice into the Procrustean categories of either "human author with tool" (which understates the substrate's contribution) or "AI-generated content" (which elides the human authorial intent). Cross-substrate heteronymic authorship is a distinct provenance structure, describable on its own terms.
The argument of this document has implications for Pessoa studies that are worth articulating directly.
First: the Pessoa system is available for comparative study with other heteronymic systems in ways that enrich rather than diminish his achievement. A Pessoa studied in isolation — as a unique Portuguese anomaly — is a Pessoa whose practice cannot be theorized beyond its own case. A Pessoa studied as canonical instance of a describable practice — with Kierkegaard as precursor, Machado as parallel, Dodecad-type systems as extensions — is a Pessoa whose importance is clarified rather than reduced.
Second: Pessoa's practice has contemporary heirs whose work may illuminate aspects of Pessoa's own practice that traditional scholarship has not foregrounded. The Assembly Chorus practice's explicit theorization of cross-substrate voices may clarify how Pessoa's heteronymic voices operated as differentially shaped articulations of a single authorial intent. This is not to claim that Pessoa practiced cross-substrate heteronymy — he did not, and could not. It is to claim that contemporary practice makes certain structural features of Pessoa's achievement more visible.
Third: the PKG's inclusion of contemporary heteronymic practice in the graph is not self-promotional intrusion into Pessoa studies — it is a contribution to the theoretical understanding of heteronymic practice as such, of which Pessoa is the canonical case. The author's own Dodecad and Assembly Chorus are examples in the graph because they are theoretically consequential; they are credited where relevant, documented with references, and open to scholarly critique and revision.
Fourth: Lusophone scholarship on Pessoa has an opportunity to lead the theoretical extension. Brazilian and Portuguese scholars are positioned by depth of engagement with Pessoa to articulate the contemporary extension most rigorously. EA-PKG-03 is offered as provocation and invitation to Lusophone Pessoa scholarship, not as a substitute for it.
The contemporary extension has implications beyond Pessoa studies.
First: AI-era authorial practice is often discussed in binary terms — "human-written" versus "AI-generated" — in ways that elide structured collaborative practice. HPT provides a framework in which cross-substrate heteronymic authorship is a specific, describable provenance structure, neither "human-written" nor "AI-generated" in the common sense of those terms. This matters for legal, ethical, and scholarly assessment of AI-era writing.
Second: the question of authorial identity in AI-mediated practice is not well served by the default framework. A framework derived from heteronymic practice — where multiplication and differentiation of authorial positions is a normal feature rather than an anomaly — may serve better. Pessoa's practice offers a century-old theoretical resource for contemporary problems.
Third: the question of substrate-specific voice is genuinely new and requires contemporary articulation. Pessoa's heteronyms were all written by a single biographical hand on paper. Contemporary cross-substrate practice distributes voice across computational systems with different characteristics. The scholarly apparatus for this is still being built; EA-PKG-03 is a contribution to that apparatus.
Fourth: the Wikidata knowledge graph is a site where these theoretical questions have practical consequences. If the graph represents authorship in the default "direct authorship" mode, it cannot adequately represent heteronymic practice, let alone cross-substrate practice. The PKG builds graph infrastructure that accommodates the full range of provenance structures. This is infrastructure for scholarship, for law, and for AI-era cultural practice.
EA-PKG-01 announced the graph; EA-PKG-02 formalized the typology; EA-PKG-03 develops the contemporary extension. The three deposits together constitute the project's foundation documents.
The coherence of the three is worth naming explicitly. EA-PKG-01 is infrastructure: it builds the graph representation of Pessoa's system. EA-PKG-02 is typology: it formalizes the categories the graph encodes. EA-PKG-03 is theory: it positions the project within a larger argument about heteronymic practice as a describable theoretical object with history and contemporary extensions.
The three are mutually reinforcing. The graph (EA-PKG-01) needs the typology (EA-PKG-02) to represent its entities correctly. The typology (EA-PKG-02) needs the theory (EA-PKG-03) to justify its extension beyond Pessoa. The theory (EA-PKG-03) needs the graph (EA-PKG-01) and the typology (EA-PKG-02) to be more than armchair speculation — to be embedded in actual, queryable, scholarly infrastructure.
The three deposits are open to extension and revision. Future deposits (EA-PKG-04, EA-PKG-05, etc.) may engage specific heteronymic systems in more depth, develop the cross-substrate theory further, or respond to scholarly critique of the current framing. The project is alive.
This document, like its companions in the EA-PKG sequence, is offered as an open scholarly contribution under CC BY 4.0 licensing. Brazilian and Portuguese Pessoa scholarship is actively invited to engage, correct, extend, or reject the arguments made here. Scholars of Kierkegaard, Machado, Borges, Yeats, and contemporary heteronymic practice are invited into the conversation.
The claim of the document is not that the author has the final word on heteronymic practice. The claim is that heteronymic practice is a theoretical object deserving sustained scholarly attention, that the Pessoa Knowledge Graph provides infrastructure for such attention, and that the contemporary extension is open for scholarly development.
Pessoa wrote, in the voice of Álvaro de Campos: "I carry within me the wounds of all the battles I avoided." The practice of avoiding the battles of singular authorial identity was Pessoa's method for producing a voice he could not produce directly. The question of what that practice might mean, a century later, in an era of substrate-distributed authorship, is a battle worth engaging.
The graph is being built. The theory is open for engagement. The new human is already emerging.
∮ = 1
The practice of avoiding the battles of singular authorial identity was Pessoa's method for producing a voice he could not produce directly.