Methodology · Con el agua de por medio

What this document declares

The atlas is a research instrument, not a visualization. This page fixes its construction rules so that any reader can audit it: follow a current on the map down to the source that sustains it, or down to the honest mark that the source has not yet been verified. The numbers here describe the corpus as of September 2026 and will change with each version; the rule is that this page changes with them.

The unit of analysis

The unit of analysis is neither the country nor the author: it is the phenomenon that runs through the basin of the Greater Caribbean. The corpus recognizes ten: coloniality, post-plantation, racial capitalism, climate catastrophes, sovereignties, violences, memory and spirituality, marronage, creolization, and body, gender and erotics. A work enters the atlas when it works one of those phenomena in depth from the Caribbean, not when it merely mentions it. The place that hosts the work is recorded, but stays in the background: Cartagena is better understood alongside New Orleans and Paramaribo than alongside Bogotá.

The admission criterion is the academic register or documentary density of the work, not its fame. The rule also excludes: in July 2026 a general-history school textbook of the Caribbean was turned away because its instructional register does not converse with the rest of the corpus. Decisions of that kind are made work by work and left on record, because a corpus is also defined by what it does not admit.

In September 2026 the academic direction opened three reasoned exceptions to that criterion, declared here. A work of fiction was admitted —Wilson Harris’s Palace of the Peacock (1960)— because it rehearses, as a novel, the dissolution of historical strata that Caribbean theory would later name creolisation and opacity; until then the corpus catalogued only works of argument. A journal article was admitted —Norman Girvan’s “Reinterpretar al Caribe” (2000)— because the “Which Caribbean?” layer needs the region’s definitions in their original formulation, and this is one of them. And two volumes of the institutional series Otras Colombias posibles (Ministry of Cultures and National Archive of Colombia) were admitted: a public-facing register, but with critical apparatus and archival work of its own, directly pertinent to the Colombian Caribbean. The three exceptions concern register, not criterion: a basin-wide phenomenon remains the condition of entry.

In October 2026 a fourth register exception was opened: testimonies and edited primary sources. Admitted were Paulette Nardal’s essays in T. Denean Sharpley-Whiting’s critical edition, the collected writings of the Guyanese activist Andaiye, a book of interviews with queer Caribbean writers and activists, and the field journals Colin Clarke kept in Kingston in 1961 and 1968. They enter for the same reason that justified the earlier ones: they document basin-wide phenomena from within, where the academic literature tells them from outside, and they come with critical apparatus. The condition of entry remains the phenomenon, not the register.

That same October the academic direction set four further rules, which this version applies. Literary criticism enters only when it reads a work or a programme the corpus already holds: hence the studies of Palace of the Peacock, of Chamoiseau and of Glissant, while a book on an absent author stayed out. Anthologies are catalogued as one volume, because the selection itself is an act of regional canon-making, and no entries are created for each text they collect. Pan-Africanism enters when its axis is Caribbean, as with Padmore, C. L. R. James or Amy Jacques Garvey, and not when the axis is Africa. And a journal article enters only if it covers a phenomenon with fewer than twenty works in the corpus. This rule has three declared exceptions. Deborah A. Thomas’s “What the Caribbean Teaches Us” (2022) entered in version 1.31.0 because it takes stock of the anthropology of the Caribbean, the very field the atlas maps. Vanessa Agard-Jones’s “What the Sands Remember” (2012) entered in version 1.33.0 because four works in the corpus cite it, one of them discussing it in the body of the text, and because it thinks queer memory and the catastrophe of Mount Pelée in Martinique together, as no catalogued book does. And Clevis Headley’s “Afro-Caribbean Philosophy” (2011), a chapter of the Oxford Handbook of World Philosophy, entered in version 1.34.0 on the same grounds as Thomas: it takes stock of Afro-Caribbean philosophy, the field of the group that sustains this atlas. By the same logic two Brazilian titles were excluded, one on the Atlantic between Rio, Recife and Africa and another on Ceará, as lying outside the basin.

Version 1.33.0 opened a fifth register exception, the literary chronicle, with a single title: Edgardo Rodríguez Juliá’s El entierro de Cortijo (1983). It enters because it is the primary source that both of Ángel G. Quintero Rivera’s works in the corpus analyse: the popular mourning for Rafael Cortijo in San Juan is the event that this sociology of music reads. As with the earlier exceptions, a basin phenomenon remains the condition of entry; the chronicle does not open the door to travel writing or journalism in general.

Also in October the entries that bundled two books under one identifier were split. Glissant now has Poétique de la Relation and Le Discours antillais; Mintz, Caribbean Transformations and Sweetness and Power; Monahan, his two books. Brathwaite, Ortiz and Benítez Rojo keep a single entry under its exact title. Each relation was reassigned to the book its source cites, and those citing only a concept —tidalectics, which has no entry of its own— became declared comparisons. It is the exact-title rule applied to the catalogue itself.

That shift from the unit of country to the unit of phenomenon has a name in the literature: archipelagic thinking. Stephens and Martínez-San Miguel define it against the area logic that has organized Caribbean studies since the Cold War, the one that groups by imperial language or national border, and propose instead a relational comparison between islands and shores, with no continental center to rank them. The atlas does not cite that framework as decoration: it is the most precise description available of what this tool already does when it connects Cartagena to New Orleans, or Providence to Curaçao.

Source: Stephens, M. and Martínez-San Miguel, Y. (eds.) (2020). Contemporary Archipelagic Thinking: Towards New Comparative Methodologies and Disciplinary Formations. Rowman & Littlefield International.

Recorded objects and epistemic operations

The corpus distinguishes between works and cultural manifestations. This difference describes the kind of object recorded by an entry; it does not separate research from ethnography. Ethnography is widely present in the corpus, but the inherited structure cannot yet identify it consistently. In architecture 2.0 it will be recorded as an epistemic operation —“producing ethnography”— that may coexist with reconstructing an archive, listening, mapping, theorizing, narrating or documenting a practice.

The manifestations layer records living expressions of the Caribbean (carnival, vodou, the sound system, the palenque) as entries in their own right, with their own nodes and relations, rather than leaving them only as objects mentioned by the books that study them. As of September 2026, it contains 0 entries against 421 works. This count does not measure how much ethnography exists in the atlas.

Relation types

The atlas works with two related but not identical vocabularies, and this revision separates them explicitly. Resonance and dissonance are the two types of relation between works, and they take up the sonic metaphor through which Michael Monahan thinks the dynamic bond between difference, recognition and freedom: freedom is creolizing because it is a relation between diverse elements that reshape each other over time, and that dynamism is incompatible with any ontology of purity or closed system. In that metaphor friction is not a flaw to correct but the physical condition of sound itself: without rub there is no vibration. «Through relations of resonance and dissonance, new possibilities are made available» (Monahan, 2023, p. 8). The atlas inherits this architecture rather than inventing its own, but it declares where each piece comes from.

Resonance is what Monahan calls constructive interference: two currents of thought that, in interacting, enhance each other without fusing into a single voice. It is not agreement or unanimity. Monahan insists that resonance needs distinction to function as resonance, the way a chord needs different notes to sound like a chord rather than unison. The atlas marks this link when one work continues, amplifies or inherits another’s: a genealogy of ideas that reinforce each other rather than repeat.

Dissonance is not the same as criticism or plain disagreement. For Monahan it is the tool that interrupts the white noise of a dominant intellectual soundscape, the imposed silence that passes for harmony or stability. Introducing a dissonant voice is often the first move of a liberatory praxis: it establishes the presence of a perspective that was previously inaudible and opens space for forms of resonance the prior order did not allow. Monahan adds a point the atlas cannot fully honor: a dissonant note can later resolve into a more complex harmony, and what a context normalizes as resonant can in fact be destructive interference wearing the mask of agreement. The atlas works with a necessary simplification of that idea: it marks dissonance when one work strains, criticizes or contradicts another explicitly and verifiably in the text. That is an operational proxy for Monahan’s category, narrower than the category itself, and the gap is declared here. And one precision this revision puts in writing: a relation established as a dissonance says nothing yet about its effects or its reach. Monahan holds that dissonant expressions are «often necessary for effective liberatory practice» (p. 89), and that often is the decisive word: dissonance can be necessary for a liberatory practice, but not every dissonance is. The atlas records the type of relation; the effect —emancipatory, transformative, or none— is analysed case by case with argument and evidence, and is never deduced from the label.

Epistemic friction is not a third type of relation, nor a category of Monahan’s. José Medina formulates it and Monahan incorporates it into his argument —«what José Medina calls epistemic friction», he writes on p. 85— within a creolizing methodology. It occurs when a framework of knowledge meets resistance that reveals its limits, silences or assumptions and opens the possibility of transforming it. In the atlas it works as a second analytical axis: it does not replace the type of relation, it is laid over it and requires evidence of its own. It is not equivalent to a difference of language. From this revision onward, the atlas separates a linguistic crossing —contact between traditions kept apart by imperial and academic borders— from epistemic friction, which requires an argued editorial claim and evidence in the relation. Until the corpus explicitly declares such frictions, the interface will neither count nor infer them.

Each category therefore answers a different control question, and answering one does not answer another:

Resonance — what idea, problem or procedure continues, answers or is transformed? A type of relation between works.
Dissonance — what argued tension or disagreement modifies the relation between the works? The other type of relation, and not in itself equivalent to emancipation.
Epistemic friction — what framework of knowledge meets resistance, and what limit is exposed? The second analytical axis; it requires evidence of its own.
Linguistic crossing — what languages or publishing traditions come into contact? An independent field, derived from the tradition labels, and no proof of friction by itself.

The current corpus records 1556 resonances and 22 dissonances. That imbalance does not describe the field alone: a field this given to polemic does not produce 22 disagreements in 1578 links. It also describes the proxy, since verifying in the text that one work cites and continues another is simpler than verifying that it deliberately strains it; the operational simplification above defaults toward resonance. The editorial task under way is to reread already-connected pairs and look in them for the real tension that the declared inheritance does not show.

Sources: Monahan, M. J. (2023). Creolizing Practices of Freedom: Recognition and Dissonance. Rowman & Littlefield. · Medina, J. (2013). The Epistemology of Resistance. Oxford University Press. Monahan’s book has its own entry in the atlas corpus: see the «Monahan» entry →

The «Which Caribbean?» layer and the drift

The atlas does not define the Caribbean. It records other people’s definitions instead: 20 formulations published between 1945 and 2026 that delimit the region in mutually incompatible ways. Each definition projects onto the same 33 atlas places one of four states —inside, at the margin, explicitly outside, in dispute— or none at all, where the text says nothing. Silence is recorded as silence and not as exclusion, because a definition is known as much by what it names as by what it leaves out. No state is inferred: each comes from a passage cited with page on that definition’s card, and the 21 disagreements between authors are entered as disputes with their note.

«The drift» is the reading device for that layer. It walks the 20 definitions in chronological order and projects them one after another onto the same map, so that what is seen moving is not the Caribbean but the disagreement of those who define it. Each place that changes state leaves behind a ring in its former colour —the wake— and a ribbon lists, step by step, what changed and from which state to which. The flicker index counts how many times each place changes state across the series: of the 33, only Jamaica never changes —it is inside all 20— while La Guajira and Palenque change sixteen times, and Cartagena, Limón, Santa Marta and Veracruz fifteen. That flicker measures theoretical disagreement rather than geographical vagueness, and it marks where the argument about what the Caribbean is actually takes place. The count depends on the order of the series and, between definitions of the same year, that order is conventional: what the index supports is not the exact figure but the hierarchy of instability. The drift adds not one datum to the corpus: it is a way of reading those already there, and the explanation of its controls lives inside the piece itself, under «How it works».

The corroboration rule

Every relation carries a source field, visible on the work’s card. A relation is only marked as corroborated when the connection has been verified against the full digital text of the work, searching for the actual citation, mention or discussion; never by title, by memory, or by intuited thematic affinity. Where that verification has not been done, the field reads «Por corroborar» (to be corroborated), and it is published that way. We prefer the marked doubt to the invented source.

Of the current 1578 relations, 1348 are corroborated and 230 remain pending. The atlas publishes that state in every version, because declared debt is information, not shame.

The pending percentage rises with each new seeding of works and falls with verification sessions. Both are fine: it means the corpus grows faster than it is audited, and that the audit is never skipped.

Workflow and roles

Every work travels the same road before appearing on the map. The book is obtained in digital form and processed in an analysis notebook under a set protocol that extracts bibliographic metadata, contribution, classification by phenomena and cited connections with page numbers. That record is systematized in a Zotero group library, where each work is a complete bibliographic item with controlled tags and annotated relations.

Through version 1.15.0, the repository synchronizes the JavaScript file loaded by the page with the downloadable JSON, but it does not yet include automated content validation. Before migration to architecture 2.0, a validator will be implemented to check identifiers, relations, vocabularies, sources, corroboration states, languages, temporalities and consistency across fields. Editorial review remains indispensable: no automated check can decide on its own whether an intellectual relation is adequately argued.

Academic direction and the editorial decisions on inclusions, exclusions and relation types belong to Roberto Almanza Hernández (Pensamiento Caribe Research Group, Universidad del Magdalena). Bibliographic systematization is carried out by the group’s research assistance team.

Open data, versions and citation

The complete corpus (works, places, phenomena and relations with their source field) is published as a dataset in JSON format, deposited in Zenodo with a persistent identifier. The concept DOI —which always resolves to the latest version— is 10.5281/zenodo.21483568; The corpus is at v1.43.1, which adds the internal register of each link without changing the public categories (v1.23.0), takes the San Andrés node from one work to five (v1.25.0) and brings in twenty-two works from the research group's library that no earlier seeding had cross-checked, among them Roberto Fernández Retamar's Calibán, Orlando Fals Borda's Historia doble de la Costa and Ochy Curiel's Un golpe de Estado on Ruling 168-13 (v1.26.0), and adds twenty more, including the first Venezuelan work, which opens the Barlovento node, and the women’s genealogy of Negritude with Paulette Nardal and T. Denean Sharpley-Whiting (v1.27.0), and splits the bundled entries and adds sixteen more works, among them the four CLACSO anthologies and Padmore’s Caribbean Pan-Africanism (v1.28.0), and a review release with no new works confirms the current of 116 works, checks 116 vaguely sourced relations against the full texts —34 are now documented and 47 become declared comparisons— and prunes the dissonances, dropping two ill-founded ones and establishing Glissant against créolité with a citation (v1.29.0). The last version deposited in Zenodo is v1.22.0 (310 works and 841 relations) and has its own version DOI, 10.5281/zenodo.22886084 (v1.4.0 added nine entries aimed at the biases this document declares, and v1.5.0 six more drawn from the group’s own library, opening the indigenous axis that had been missing, with Whitehead and Rifkin; v1.6.0 and v1.7.0 seeded the Caribbean Colombia/independence cluster —Helg, Brathwaite, Múnera and Sæther— with several relations corroborated by verified cross-citation between the primary texts themselves, including a genuine dissonance: Sæther directly challenges Múnera’s thesis on Cartagena’s free people of color; v1.8.0 added a chapter by the researcher himself —Almanza and Granados Castellanos, on the black geographies of the Zona Bananera— published in Raíces, resistencias y territorios (Editorial Unimagdalena, 2026); v1.9.0 added El cobarde no hace historia, Joanne Rappaport’s study of Orlando Fals Borda and participatory action research in the Sabanas of the Colombian Caribbean coast (Editorial Universidad del Rosario, 2021); v1.10.0 read both texts in full and found verifiable direct citation: the Zona Bananera chapter engages Trouillot, Bassi, Glissant, Abello and Rediker, and Rappaport’s book builds its analysis of Fals Borda’s method around Trouillot’s historiographic framework; v1.11.0 added two more works —Cristo Rey, un espacio para permanecer en el tiempo, by Édgar Rey Sinning, and Culture et dictature en Haïti, by Laënnec Hurbon— and, on reading the first work’s full draft, upgraded two relations from "to be corroborated" to verified citation and added three more, with Escalante, Helg and Cunin; v1.12.0 added three works drawn from the group’s own library: Leyendo el carnaval (Lizcano Angarita and González Cueto, eds., 2009), the first work to anchor the atlas in Barranquilla, previously at zero; Volviendo al Caribe (Mukien Adriana Sang Ben, 2020), which cites at length both already-catalogued works by Glissant and credits Hurbon as a reading that changed her understanding of Haiti; and ¿Dónde hay música africana? (Chica Geliz and Fajardo Martínez, 2024), on Cartagena’s picó culture and the International Caribbean Music Festival; v1.13.0 added fifteen works identified through a systematic cross-check against the group’s bibliography folder: among them Conscripts of Modernity by David Scott and Mosquito Empires by J. R. McNeill, two titles marked “not yet obtained” since the first round of recommendations; Raíces, Resistencias y Territorios, a compilation on Santa Marta that turned out to share authorship with two works already in the corpus; and Contemporary Archipelagic Thinking, the volume that theorizes the method the atlas was already practicing; v1.14.0 added nineteen works identified through a second systematic cross-check against the group’s bibliography folder, weighted toward Panama —until then almost entirely absent from the corpus, with Lasso, Flores-Villalobos and Corinealdi citing one another— and toward Haiti, Cuba and the Dominican Republic; among them also The Argonauts of West Caribbean by Wilhelm Londoño Díaz, on the canoe tradition of Taganga, and The People of Puerto Rico, the 1956 collective project that produced the already-catalogued Sidney Mintz. v1.15.0 added thirty works from a third systematic cross-check against the group’s bibliography folder, shifting the weight of the corpus more than any earlier round: it brought in the classics the atlas had been citing second-hand without holding them —Eric Williams’s Capitalism and Slavery, Moreno Fraginals’s El ingenio, M. G. Smith’s plural society, Sylvia Wynter’s unpublished Black Metamorphosis, Edith Clarke’s My Mother Who Fathered Me and the essay Mintz and Price wrote together—, opened two new nodes, Costa Rica and Grenada, and seeded the Indian and Chinese indenture axis in one pass. Of its 159 relations, 149 carry a citation verified in the copy, raising the corroborated share of the corpus from 54% to 65%. v1.22.0 is the last deposited version, on September 22, 2026; from that version the deposit also includes the canonical .js source and the synchronization script, plus the corpus validator — not the Atlas 2.0 pilot validator, which depends on files still in transition and could not run from the deposit. v1.16.0 merged two entries that turned out to be the same 1956 book, The People of Puerto Rico, so the living corpus had one entry fewer than the deposited one; the retired identifier redirects to the surviving one so that already-cited links do not break. v1.17.0 added twenty-five works identified through a fourth systematic cross-check against the group’s bibliography folder, with twelve new relations verified by exact citation against the existing corpus and among the new works. Each work in the atlas also has its own citable link and a button that copies its citation in APA format.

Declared limitations

Dissonances are underrepresented: 22 of 1578 relations. Part of the bias is the field’s, part is the operational proxy for Monahan’s category: it is easier to verify inheritance in a text than deliberate tension.
The inherited structure cannot systematically identify the many ethnographies in the corpus. Migration 2.0 will add “producing ethnography” as an epistemic operation, separate from entry type.
Two phenomena remain thin relative to the rest: climate catastrophes (16 works) and violences (31). Targeted seeding continues there.
35.8% of the relations are pending verification against text.

None of these limitations is hidden in the interface: the state of the corpus is published in the atlas itself. This is exactly where a digital humanities project parts ways with a pretty visualization.