I present the results of mid-distance reading that combines graph and data mining methods to identify genre in narrative works written in Spanish. In this case, I analyze ten historical novels from the “National Episodes” of Benito Pérez Galdós (1843-1920). In 2003, Alex Woloch stated that narrative structures are formed by a character-system or “the arrangement of multiple and differentiated character-spaces [, that is,] differentiated configurations and manipulations of the human figure” (14). For its potential as a quantifiable study, several digital humanists such as Elson et al. (2010), Sack (2011), Agarval et al. (2012), or Moretti (2011-2016) have compared Woloch’s concept of character-system to the notion of network, where the nodes are the characters and the edges represent interactions between them.
In the present methodology I follow said notion, and I proceed in three steps to curate and analyze data: 1) One algorithm (written in R) detects names in the texts and substitutes them by a resolved name from a file of metadata of 4,644 characters; 2) Another R algorithm extracts the number of co-occurrences of characters by chapter; and 3) I calculate the centrality and modularity of the sum of co-occurrences by novel in Gephi and I visualize each character system as a network. The results of this mid-distance reading show that there is a reciprocity between the positioning of the historical and fictional characters in the character-systems and the typology of the novels that had been previously identified by the close-reading of galdosistas such as Urey (1989) or García Castañeda (2012). In fact, there is a greater emphasis on what Galdós called “integral history” in those novels that tell the life trajectory of the main character like a serial bildungsroman. Overall, this methodology seeks a better understanding of the possibilities that digital tools offer for the study of textual cultural operators in Spanish.