From print to digital: A web-edition of Giacomo Leopardi’s Idilli

Desmond Schmidt (, Queensland University of Technology, Australia and Paola Italia (, Università di Bologna, Italy and Milena Giuffrida (, Università degli studi di Catania, Italy and Simone Nieddu (, La Sapienza, Università di Roma, Italy

Giacomo Leopardi (1798-1837) was a significant Italian romantic poet best known for a volume of poetry, Canti , published in 1835 in Naples, one copy of which he manually corrected shortly before his death. Among his poems, the series entitled Idilli ‘Idylls’ (1819-1821) comprises Alla Luna, L’infinito, Lo spavento notturno, La sera del giorno festivo, Il sogno and La vita solitaria . In the earliest extant form in the Naples Notebook the poems were written in three separate phases (Idylls 1-3, 4-5 and 6), then corrected by the author in identifiably different pens , before being copied into the Visso manuscript, where they were again revised. Their first publication was in the review ‘Nuovo Ricoglitore’ in 1825/1826, then in the Bologna edition of Versi ( 1826). Except for the third Idyll: Lo spavento notturno , the other poems were collected in the Florence edition of the Canti ( 1831), then reprinted in the Naples edition (1835), in which Lo spavento notturno appears as one of the Fragments (n. XXXVII).

This textual history of Leopardi’s Idilli is interesting because much of Italian Philology of the Author ( Variantistica ) has been based upon it. After Francesco Moroncini’s first critical edition of the Idilli (1927), three other editions have appeared: Peruzzi (1981), De Robertis (1984) and the latest by Franco Gavazzeni, printed by Accademia della Crusca ( Leopardi , 2009). This is both a critical edition of the manuscripts and the print-versions. It contains detailed textual notes recording all handwritten changes including marginal alternative readings, which are characteristic of Leopardi’s method of correction. Four different pens have been identified in ‘Naples notebook’ of the Idilli, before their copying into the Visso manuscript: A, B, C and D, which were used both for writing the new texts and for correcting the earlier idylls. For example, Leopardi wrote idylls 4 and 5, then corrected idylls 1, 2 and 3. With pen C he then wrote idyll 6, and then finally corrected all the preceding ones.

A wiki edition of Leopardi was also produced as part of an advanced course in Italian Literature in 2016-2017 at the University of Rome La Sapienza (Giuffrida and Nieddu, 2017; Caterino and Nieddu, 2017) and was based on Gavazzeni’s critical print edition of the Canti.

The technical limitations of the wiki software used (MediaWiki), led the group of editors to seek better ways to encode and present the text. They were also interested in producing editions of other authors, and to have the capacity to expand the Leopardi edition in future. As a result of a lecture given by Paul Eggert in October 2017 at the University of Bologna, Eggert suggested the use of the Ecdosis editing system which had already been developed for the Australian romantic poet Charles Harpur. Other tools, such as Tapas (Bauman et al., 2017), EVT (Rosselli Del Turco et al., 2014), CollateX (Dekker and Middell, 2017) and Juxta (n.d.) had also been considered but in spite of their individual strengths none provided the comprehensive web-based editing system for modern manuscripts that the editors were seeking.

Leopardi’s Idylls were chosen as the basis for the pilot project. Although it was designed as a general set of web-tools, this was the first time that Ecdosis had been applied to another editorial project. Conversion of the print and wiki editions into Ecdosis took only 12 days of part-time work. Since Ecdosis’s own WYSIWYG editor was still incomplete it was decided to encode the text in XML and import it. The XML files were created by copying from the wiki edition and from a PDF of the print edition (Italia, 2018). The detailed textual notes of the print edition were used to create separate files for each of the identifiable versions. Within each version changes were encoded with the usual and codes and our own code for earlier alternatives that were not cancelled. The importation process split the corrections and their contexts into separate layers, amalgamating local levels of correction into coherent sub-versions. Although these layers were never written by the author they are still useful as a storage mechanism to record local changes within a version. In this way individual layers remain simple, needing only a few codes to denote changes in format, such as lines, headings and stanzas, since all deletions and insertions have already been converted into layers. Figure 1 shows an example of how this process works for a segment of Idyll 3, La Luna, from the Naples notebook. Finally, the separated versions and layers of each poem were stripped of their remaining markup, which in Ecdosis is stored separately and only recombined with the text for display, so reducing each version/layer to a readable plain text file. This greatly simplifies all subsequent text processing such as searching, comparing and hyphenating when compared to the complexity of the original XML.

Hyphenation of XML encoded texts is quite difficult due to the variety of tags that may occur between two halves of a word. (Bauman, 2016).Our tests of search engines on major digital scholarly editions revealed that literal searches (often more useful than keyword searches) do not work across internal variant boundaries (,, etc). Comparison between TEI-XML files containing inline variants is still an open problem that requires human intervention (Bleeker, 2017).

Figure 1: Falsely colourised portion of Naples notebook showing pen A (red) and pen B (blue)

During importation the page-images were also linked to the text. The manuscript images are copyright to the National Library, Naples, but it is anticipated that permission to publish these with the edition will be obtained soon. For the moment the site is protected by the same passsword as for the wiki. The linking produces a list of images which scrolls in sync with the text to keep the top and bottom of each page-image aligned with its corresponding position in the text. To provide a smooth transition and accurate alignment between text and image when scrolling all other possible alignment positions are calculated in proportion to these fixed points. Changing the version loads the relevant images on the left hand side of the screen and the corresponding text on the right. Layers within a version can be changed by clicking on a tab above the text, and changes between layers are highlighted. This removes altogether the need for a ‘diplomatic’ display where changes are displayed awkwardly above or below the line using inline formats.

In Compare View differences between versions and layers are shown at the character level: deletions on the left-hand version/layer in red and additions on the right-hand version/layer in blue. When displaying a layer the invariant text is shown in grey to indicate that this is not a true version, but true versions and final layers are displayed in black. Scrolling is also synchronous and aligned left to right, regardless of differences in length between the two currently displayed versions.

Table View resembles the traditional critical apparatus. Differences and similarities between versions/layers are arranged in columns. Versions/layers can be excluded from comparison and the table rebuilt. Also versions can be moved up or down the display to explore specific clusters of variation for editorial purposes. The establishment of a reading text from this information could be encoded as another version and added to each poem as a default text.

The pilot edition has mostly been a success (Leopardi, 2018). There is still some difficulties with the encoding of marginal alternatives which cannot be placed with certainty in the text. These will probably be encoded as annotations instead. Another problem is that the modules in Ecdosis and the website itself currently resemble too closely those of the Charles Harpur Critical Archive (Eggert, 2018) and will therefore need significant customisation. The possibility to export the entire contents of the edition to a simple collection of files in nested folders, encoded in standard HTML and plain text, has mitigated initial concerns about ‘lock-in’. It is hoped to use other Ecdosis tools to expand the edition in future by increasing the number of poems and placing them in the context of research into the issues and people of Leopardi’s day.

Appendix A

  1. Bauman, S. (2016). The Hard Edges of Soft Hyphens. Accessed 2 May 2018).
  2. Bauman, S., Clark, A., Quinn, B, Flanders, J., Hamlin, S. and Zoller, E. (2017). Tapas Project. (accessed 30 April 2018).
  3. Bleeker, E. (2017). Scholarly Intervention in Automated Collation Software, ESTS Book of Abstracts 2017 (accessed 2 May 2018).
  4. Caterino, M. and Nieddu, S. (2017). Wiki Leopardi. , user: wiki_leop, password: leopardi (accessed 30 April 2018).
  5. Dekker, R. and Miiddell, G. (2017). CollateX – Software for Collating Textual Sources. (accessed 30 April 2018).
  6. Eggert, P. (2018). Charles Harpur Critical Archive, (accessed 30 April 2018).
  7. Italia, P. (2008). I tre tempi degli «Idilli» Leopardiani (con un’edizione del quaderno napoletano), Filologia Italiana, 4, 173-213.
  8. Leopardi, G. (2009). Canti e Poesie disperse, edizione diretta da Franco Gavazzeni, edited by Paola Italia, Florence: l’Accademia della Crusca.
  9. Giuffrida, M. and Nieddu, S. (2017). Wiki Critical Editions: a sustainable philology, in AIUCD 2017 Conference, 23-28 January, pp. 257-258. (accessed 30 April 2018).
  10. Rosselli Del Turco, R., Buomprisco, G., Di Pietro, C., Kenny, J., Masotti, R. and Pugliese, J. (2014). Edition Visualization Technology: A Simple Tool to Visualize TEI-based Digital Editions, JTEI 8. (accessed 30 April 2018).
  11. Leopardi, G. (2017). Idilli di Giacomo Leopardi. (accessed 30 April 2018).