Giacomo Leopardi's Zibaldone as a Digital Research Platform
Edited by Silvia Stoyanova and Ben Johnston

Project Introduction

The project of building a Digital Research Platform for Giacomo Leopardi’s Zibaldone of Thoughts aims to create a computational model for the semantic organization of this voluminous collection of research fragments and to design an interface that would provide users with affordances to explore and configure the relationality of its networked textuality, engage in its performative interpretation through semantic queries, dynamic visualizations and remix, augment the authorial semantic annotations via participatory modules, share research in machine-readable format and contribute to the editorial apparatus. The project is premised on the hypothesis that Leopardi's various semantic pointers—marginal additions, cross-referencing, thematic indexing, intertextual references—would gain discursive articulation in the digital medium, and that their computational processing would give insight into his epistemology. The editorial approach of remediating the Zibaldone as a digital research platform responds, on the one hand, to the reader's task of reconstructing the semantic framework of the fragments and, on the other, corresponds to the function the Zibaldone had for its author, namely a personal knowledge base whose paper manuscript technology was inadequate for mediating his phenomenological research methods into a form of discourse that would sustain and potentiate the signifying agency of observed phenomena. The Zibaldone project thus fundamentally explores qualitative methods of employing digital technology to support the process of interpretation and reflects on the ethical, aesthetic and mediatic criteria of constructing the unity of discourse in humanities scholarship.

The Zibaldone Genre. A private collection of 4526 manuscript pages of research notes and observations on human culture, divided by their date of composition, the bulk of which was gathered over the course of ten years, written in several languages, citing thousands of intertexts, copiously annotated with marginalia and cross-references connecting related fragments, thematically indexed by the author at the paragraph level, the Zibaldone is not a formally finished work and a very small part of its contents found its way into Leopardi’s publications. The Zibaldone's textual genre and authorial annotations of semantic analysis can be productively compared to a personal knowledge base—a technology affording its users to capture, store, retrieve and express their acquired knowledge, whose form can range from the subjective recording of an idea to the identification of a relationship between two sources in the objective realm (Davies et al. 2005; Davies 2011). At the same time, the Zibaldone shares little more than its name with the early modern predecessor of the personal knowledge base, i.e. the commonplace book or zibaldone: the excerpts and bibliographic references amount to less than a tenth of its textual contents. Leopardi's collection of fragments, rather, belongs to the strand of intellectual notebooks that became custodians of the author’s unrealized opus: the Carnets of Joseph Joubert, the Notebooks of Coleridge, Novalis’ Allgemeine Brouillon, Valéry’s Cahiers, Benjamin’s Passagenwerk, among others, are kindred texts whose “infinitely resonant interconnectedness or relationality” (Gifford and Stimpson, 1998: 303) reflects an encyclopedic movement of thought in the romanticist sense of encircling a phenomenon from multiple angles. Fragment writing is engendered by the epistemological method of phenomenology—“the taking up of a certain attitude and practicing a certain attentive awareness to the things of the world as we live them rather than as we conceptualize or theorize them”—and its tension between the experience of the "presentation" of phenomena and their verbal "representation" (van Manen 2006: 720; 718). The discursive efforts of these notebook authors to convey the relational structure of their perception of phenomenal reality tend to fragment the constructs of language and of the manuscript page, and turn to graphical configurations and three-dimensional expedients to capture its semantics: coss-references, marginalia, and thematic tagging on index cards in Leopardi’s Zibaldone; “linking loops and marginal additions […] classificatory tables of all sorts, graphics, doodles […] sumptuous water-colors” in Valery’s Cahiers (Gifford, 1998: 38); “syntax [which] transforms itself into a diagrammatic disposition” in Joubert’s Carnets (Kinloch, 1996: 349); cross-referencing, layouts, tables, colors, diagrams, envelopes, folders, cutting up and re-arranging slips of paper, etc. in Benjamin’s Arcades Project and Notebooks (Marx et al., 2007), etc. These vast fragmented corpora of the intellectual notebook genre necessitate the collective editorial agency of a community of readers and the computational, hermeneutic and poietic agency of digital technology to build the dimensional discourse constituted by their phenomenological attunement, profoundly relativist perspective, encyclopedic ambition, and continuously suspended telic intentionality. Conversely, the computational modeling and digital articulation of their semantic discourse could assist the deconstruction of the enduring print paradigm embedded in humanities scholarship and the development of fluid, processual, relational, distributed, performative, multi-modal environments for scholarly work, as envisioned by some digital humanists and media theorists.

Modularity and Relationality. Whereas the relationality of research fragment collections is flattened and obscured by their paper technology, they are well-suited for digital remediation because their textuality is inherently modular and, in the case of Leopardi's Zibaldone in particular, the semantics of the fragments is painstakingly analyzed and numerically coordinated—by date divisions serving as semantic markers, by cross-references semantically linking textual nodes, by thematic indexing at the paragraph level, and with precise bibliographic references to the editions of texts cited and consulted. The logistical challenges that readers of the Zibaldone have to confront in the attempt to define any given subject of inquiry and account for all its ramifications, originate in this very modularity that is generated by the authorial perspective—distributed over an extensive period of time and refracted through multiple contexts, as well as in the polysemy of the fragments gained in this iterative process of composition and also replicated at the syntactic and lexical levels of the text. With the intention to sort the recollected material thematically for the publication of scholarly works, Leopardi furnished his collection of fragments with an alphabetical thematic index (the 1827 Index) and with the titles of projected works (the PNR Index), listing altogether over 10 000 references to relevant paragraphs and pages, distributed under ca. 1000 themes and sub-themes, some of which are cross-referenced. The thematic indexing combined with the cross-references within the Zibaldone sketch a detailed conceptual design of the semantic framework of the text, where each fragment is connected to multiple others at various degrees of association.

Leopardi’s analytical procedures probe extensively the semantic associations between individual fragments, however their relational structure cannot be retrieved from the chronological order of the collection or from the arbitrary alphabetical order of the index headings, which Leopardi had adopted from his eighteenth century "reasoned dictionary" models of knowledge organization, i.e. Voltaire’s Dictionnaire Philosophique and the Encyclopédie ou Dictionnaire raisonné des sciences, des arts et des métiers edited by Diderot and D’Alembert. The proto-hypertextual expedient of the cross-reference, to which both the Zibaldone and the Encyclopédie resort in the endeavor to convey their relativist epistemological perspective, can activate the associative network of the fragments in the mind of their dedicated reader, who may on occasion grasp, as a flash-like image ("un colpo d'occhio", "un coup d'oeil"), how its various parts hang together, after having examined them carefully one at a time (Zibaldone pp.3269-70). However, even when remediated in digital space as hyperlinks, the cross-references lack the hermeneutical and rhetorical capacity to translate this conceptual image into discourse. Although the fabric of the Zibaldone is hypertextual, it is not a hypertext insofar as the author does not have mastery over its system of relations but, rather on the contrary, employs the cross-references to chart semantic trajectories through the fragments in the quest for their order of discourse. The task of retrieving this order requires the computational encoding and processing of the manuscript's semantic annotations. This would allow to extract the fragments from the continuum of phenomenological description and rearrange them in the dimensional space of the digital medium according to how their semantic agency is deployed in the networks of associations, with visualization graphics simulating and arresting the fleeting synoptic image of the mind, so that its constituents can be re-examined and apprehended anew.

Semantic Networks. The aggregation and retrieval of the information recorded in the Indexes and by the cross-references for the purpose of evaluating the network of relevant passages of a thematic field or even of a single fragment, demand a labor-intensive sequence of manual steps which the Digital Platform's further development should be able to render automatic. For example, if we set out to explore the smallest theme from the PNR index, "Handbook of practical philosophy", there would be 82 references to paragraphs and 6 references to themes in the 1827 Index to consider. Taking the first passage, par. 43,2 as an entry point into the maze of the Zibaldone, although it is not directly linked to other passages by cross-references, it is also listed under the 1827 Index theme "Time. Use of time”, which information is given in the paragraph information window provided by the Digital Platform, but otherwise can be found only in the index tabulations of Peruzzi's ten-volume facsimile edition. Since paragraph 43,2 contains a single sentence, this more specific sub-theme should be quite relevant, and so it is important to consider any other passages listed under it and check whether Leopardi has included them in the main topic. “Time. Use of time” lists two additional passages plus a cross-reference to the 1827 Index theme "Negligence, Inactivity. Diligence, Activity"—a compound heading containing its antonym. One of these two passages is already listed under the main theme of “Handbook of practical philosophy” as well as under the cross-referenced theme of "Negligence, etc"; the other is not and should be retrieved to evaluate its relevance. “Negligence”, in turn, has four passages, one of which is listed under the sub-theme of “Time. Use of time” and one which already belongs to our main PNR theme; the other two passages then need to be retrieved and their relevance evaluated. In addition, “Negligence” has a sub-theme listing two passages, one of which repeats under the general heading of “Negligence”, however both are already listed under the main PNR theme, confirming that this cross-referenced theme is indeed relevant to the main subject. We should also consider the incoming and outgoing cross-references of all of these additional passages. Then we would have to follow similar iterations of this process for the rest of the 81 passages and the 6 referenced themes from the 1827 Index that Leopardi lists under "Handbook of practical philosophy", which would make the sorting, let alone the discursive articulation, of the related fragments increasingly complex. 

In order to follow the connections between a thematic selection of fragments on the basis of Leopardi's Indexes and cross-references, account for their repetitions, frequencies, inter-relatedness and retrieve additional passages to evaluate their relevance, one would need to go back and forth through the Indexes and from the Indexes to the Zibaldone, record which passages are relevant and in what respect, copy the text of those passages and try to arrange them in an order that reflects their relevance to the main theme and to each other, which quickly becomes unfeasible even with a digital transcription of the text. The computational processing of this information would redistribute the reader's cognitive resources from the retrieval and sorting to the evaluation and further defining of the semantic associations suggested by the authorial annotations and by the editorial criteria of the encoding. The reconstruction of Leopardi's analytical conceptualization of the semantic structure of his Zibaldone, especially by means of computer-generated Visualizations, would assist the definition of parameters and the design of affordances for its collective analysis and interpretation, as well as provide insight into Leopardi's epistemic methods. The authorial semantic analysis and the data from user-generated analysis of the fragments' semantic structure could then be translated into algorithms and compared with the results of methods for automated semantic network retrieval of unstructured text and contribute to their evaluation and development.

Platform Development. The Digital Research Platform aims to address the limitations of existing remediations of the Zibaldone, i.e. print editions, digital transcription, CD-ROM, wiki page, to confront the manuscript’s distributed semantics, by taking advantage of XML and RDF technologies which allow to critically address the Zibaldone from the perspective of its modular and relational textuality. The evolving development of the Platform explores a methodological dialectics between the Zibaldone's discursive gestures of transcending its paper medium and the prerogative of digital technologies to construct multi-dimensional forms of discourse, and is informed by methodologies of digital scholarly editing and knowledge visualization, the principles of the semantic web, the history of scholarly hypertext, and the design of platforms for the organization of research material. 

Editorial apparatus. The Platform development builds upon and remediates features of the critical apparatus of editions and other scholarship on the text with the objective to organize this scholarship relationally, aggregating multiple editorial perspectives and making them accessible modularly, scaled according to individual reader needs. Print editions and existing digital remediations provide valuable individual contributions to the analysis of the text, but they do not implement its relational features beyond basic hyperlinking (wiki project), text search with logical operators and by date and area of the manuscript (CD-ROM, Ballerini & Ceragioli 2009), and index tabulations (Peruzzi 1989-1994). The CD-ROM does not activate the cross-references as hyperlinks nor connects the indexes to the text, whereas the wiki project connects them only in one direction, from their list of references—you can jump to the relevant fragment in the text from a given index heading but you cannot retrieve all of the index headings relevant to that fragment. Hyperlinks facilitate the reader’s movement through the sequence of related fragments, however the chunk-style sequential relations they establish do not qualify their semantic values, nor build an overview of the thematic fields traversed by that sequence. The apparatus of print editions altogether tends to be accessible only in one direction and to address one query at a time. Pacella’s edition (1991), for example, identifies the dates of some marginal annotations, however this information can be retrieved only from the location of a particular annotation in the page sequence of the text; the reader cannot retrieve the annotations written in a specific time period, nor identify those which have been dated, without perusing the edition’s entire critical apparatus. The index tabulations of Peruzzi’s facsimile edition make an exception in calculating the information of the indexes in two directions: the first tabulation lists in chronological order all pages and paragraphs with all of their corresponding themes, and the second tabulation gives in alphabetical order the index headings along with the additional headings of the fragments listed under them, also in alphabetical order, with each referenced page or paragraph number repeated as many times as the number of cross-headings under which it is listed. This linear unraveling of Leopardi’s exponential method of cross-referencing and its extension over several hundred pages render concretely perceptible the need to give a dimensional structure to the relational organization of the text’s analytical procedures, as well as to the organization of its editorial commentary.

Technologies and Functionality. Our editorial efforts thus far have been dedicated to document analysis and encoding of the manuscript’s semantic annotations in XML; reconstructing the Zibaldone’s intertextual networks based on quotations and bibliographic references; and providing basic tools to query, extract, and align the text’s semantic layers. The encoding of the text and its authorial indexes according to the guidelines of TEI P5 defines the Zibaldone's modular features, with the occasional need to circumvent the TEI OHCO model. The eXist XML server allows for XQuery processing and XSLT transformations of the files and offers Lucene full-text searching and a built-in web server. While the eXist XML database provides a full application framework, Drupal was used for the website interface. With the addition of a few plug-in modules, the versatile Drupal content management system simplifies the addition of existing and future features, such as authentication, authorization, commenting, tagging, caching, and theming to the website. The Platform's current functions include: settings which give users the option to display or hide specific layers of authorial and editorial markup according to their individual research objectives; hyperlinking the intratextual cross-references and the references of the Indexes to the text, which required defining the relevant textual segment for each reference; several navigation menus; a search menu; a paragraph information window aggregating all directly referenced passages (outgoing, incoming and reciprocal) along with the themes from the 1827 Index and the PNR Index which reference the given paragraph; extracting the text of the passages referenced under an index theme and visualizing their thematic cross-references as a network graph. Among the Platform’s desiderata are a sophisticated concordance search, dynamic semantic queries and visualizations of the text's syntactic features and of its relational networks, modules for user annotation, a database of user-generated semantic indexing, links to a timeline of Leopardi's other writings, a database of Leopardi's readings, an indexed critical bibliography on the text, and other contextual information.

For a detailed background of the project's research objectives and the technologies employed, see the Editorial History and the project Publications. For information on the current status of the Platform functionality, see the User Guide. The consultation of existing editions of the Zibaldone and scholarly sources for the document analysis and encoding of the text has been documented in the project's Editorial History.

Ballerini, Monica and Ceragioli, Fiorenza, Eds. Giacomo Leopardi, Zibaldone, CD-ROM. Zanichelli: Bologna, 2009.
Davies, Stephen. “Still Building the Memex”. Communications of the ACM. February 2011, vol. 54, no. 2, pp.80-88. doi:10.1145/1897816.1897840.
Davies, Stephen et al. "Building the Memex Sixty Years Later: Trends and Directions in Peronsal Knowledge Bases. Department of Computer Science, University of Colorda, Technical Report CU-CS-997-05, August 2005. http://www.cs.colorado.edu/department/publications/reports/docs/CU-CS-997-05.pdf

Gifford, Paul. “Thinking-Writing games of the ‘Cahiers’”. In Reading Paul Valéry: Universe in Mind, Eds. Brian Stimpson and Paul Gifford, Cambridge University Press, 1998, pp.35-52.
Gifford, Paul and Brian Stimpson. Reading Paul Valéry: Universe in Mind, “Conclusion”,Cambridge University Press, 1998, pp.297-305.
Kinloch, David. Reading and Writing in Joubert's "Carnets"Author(s): The Modern Language Review, Vol. 91, No. 2 (Apr., 1996), pp. 342-354.
Marx, Ursula et al. Walter Benjamin's Archive: Images, Texts, Signs, London and New York: Verso, 2007, Kindle edition.
Pacella, Giuseppe, Ed. Giacomo Leopardi, Zibaldone. Milano: Garzanti, 1991.
Peruzzi, Emilio, Ed. Giacomo Leopardi, Zibaldone di Pensieri, edizione fotografica dell’autografo con gli indici e lo schedario. Scuola Normale Superiore: Pisa, 1989-1994.
van Manen, Max. “Writing Qualitatively, or the Demands of Writing”. Qualitative Health Research 16, 5 (2006): 713-722. 

Participants and Support

The Zibaldone project was initiated by Silvia Stoyanova (PhD in Italian Literature, Columbia University) and developed in collaboration with Ben Johnston (Senior Educational Technologist, Princeton University). The project has benefited from the consultant assistance of Dr. Clifford Wulfman (Center for Digital Humanities, Princeton University) and Prof. Christian Wildberg (Classics, Princeton University); from the editorial assistance of Stephen Blair, Emilio Capettini, Michael Hanley, Kathleen Galeano, Monica Gordillo, Gigi Stoyanova, Francesco Annibali; and from the computational programming assistance of Matthias Schneider, Mariona Coll Ardanuy, and Guoqian Xi. The project development has received support from Princeton University, the Trier Center for Digital Humanities, the University of Macerata, and volunteer work. The editor is thankful for the assistance of the staff at the Manuscripts and Rare Books Department at the National Library in Naples in July of 2014 – Emilia Ambra, Vincenzo Boni, Maria Rascaglia, Gabriella Mansi, and for the feedback of Prof. Laura Melosi (University of Macerata) and the students enrolled in her Italian Literature course in the Fall of 2016.

Feedback / Volunteer

If you have comments, suggestions, would like to report any mistakes, or would like to volunteer on the project as editor or developer, please email Silvia Stoyanova (sms116 [at] caa [dot] columbia [dot] edu).


Giacomo Leopardi’s Zibaldone as a digital research platform: a methodological proposal for its semantic reconstruction and discursive mediation. Silvia Stoyanova. In Semicerchio: rivista di poesia comparata, LIII, 02/2016, pp. 98-106, Ed. Francesco Stella.

Remediating Giacomo Leopardi’s Zibaldone: Hypertextual Semantic Networks in the Scholarly Archive. Silvia Stoyanova and Ben Johnston. In Proceedings of AIUCD ‘14, Bologna, Italy, ACM (08/2015). Ed. Francesca Tomasi et al. Alternative access.

"The Zibaldone Hypertext Research Platform". Silvia Stoyanova and Ben Johnston. In Authoring Software/Narrabase (07/2014), Ed. Judy Malloy

"Fragmentary Narrative and the Formation of Pre-Digital Scholarly Hypertextuality: G. Leopardi’s Zibaldone and its hypertext rendition". Silvia Stoyanova. In Proceedings of Hypertext 2013, ACM (05/2013)

“Lo Zibaldone di pensieri di Leopardi: progetto di una piattaforma di ricerca ipertestuale”. (“Leopardi’s Zibaldone of thoughts: a project for a hypertext research platform”) Silvia Stoyanova. In Lo «Zibaldone» di Leopardi come ipertesto, Atti del Convegno internazionale (Barcellona, 26-27 ottobre 2012). Ed. María de las Nieves Muñiz Muñiz, Florence, Leo Olschki, 2013, pp.333-342.