The Collectif interuniversitaire d’étude du littéraire (CIEL) database is the result of a collective research initiative: a collaboration between the Université libre of Brussels and the Université de Liège; it was directed by Paul Aron and Jean-Marie Klinkenberg and coordinated by Benoît Denis. The project was financed by the Communauté française de Belgique from 2002 to 2007 in Liège and Brussels, and then from 2008 to 2013 in Brussels. Several theses have resulted from it.1
The database was developed by the Centre informatique de philosophie et lettres (CIPL), at that time under the management of Gérald Purnelle.2 The CIPL, created in 1983 at the Université de Liège, is a Unité décentralisée d’informatique (UDI), which currently corresponds to a research and development centre for digital humanities. Its general objective is to promote and coordinate the use of computer science within the Faculté de philosophie et lettres of the Université de Liège and to ensure the support and development of data processing projects for research and teaching.
Luc Desert, a computer scientist at the CIPL, developed the CIEL database from 2002 to 2005. A trainee of Desert, Björn-Olav Dozo ensured (with Daphné de Marneffe) the supervision of students assigned to complete the database, but also serially ran the data.
The database has two primary goals: the first one is patrimonial, the second is more “experimental.” The first ensues from the acknowledgement that Belgian Francophone literature came about as a result of diverse initiatives thus causing major disparities in the documentation that is currently available. The database conception firstly aims to compile the already existing data, correct it if necessary, and then update it. The data processing medium is the ideal format for this type of constantly evolving data. The second goal is to statistically process this mass of data.
The database currently groups together all prosopographical data available in various existing works, such as the Biographie nationale (Thiry, then Bruylants and Christophe then Bruylants [eds.] 1866-1986) the Nouvelle Biographie nationale (1988), the Bibliographie des écrivains francophones belges (Brucher then Detemmerman [eds.] 1958-1988), and the Dictionnaire des œuvres (Nachtergaele & Trousson 1988; Berg & Frickx 1988; De Grève, d’Heur, Pouilliart 1989; Frickx R. et alii 1994). Many monographs about Belgian literature were systematically examined, as well as professional directories (Mayeur 2010),3 lists of members, and signatories of manifestos and of other collective texts. Birth, death, and place of residence registers were the object of first-hand inquiries, primarily in Brussels and Liège, but also from time to time in other Belgian cities. This data is structured and linked logically. The database collects factual data, unanalysed and formatted as little as possible, which is to say that the data intake is divided into two steps: a primary intake conforms to the source, then a second, complementary intake for analytical purposes.
When we speak about databases in literature, we often think of major digital literary corpora such as “Frantext” or, for Belgium, “Beltext.” These databases are set up mostly by linguists. The history of literature makes little use of them, even though things have evolved, notably with the perspectives opened up by the work of Franco Moretti (2005). Alongside these textual databases, the history of the literary and sociological life of authors has developed its own databases. The CIEL database constitutes a pioneering example of this.
Which software programs were used to build the database infrastructure and, as the case may be, to treat the data statistically?
The core of the database was developed using Microsoft Access, because it is easy to use, and because it is easy to make changes quickly during the preliminary development phases (i.e. going back and forth from the researchers who were going through the sources, to the computer technician, with a researcher playing the role of interface between the two). The ramp-up went well: on the university’s local network, MS Access allowed simultaneous encoding within the database by up to twenty people at the same time, via specific forms.
However, when it was necessary to capture information from a distance via the internet – with several people going through sources at the library, for example – different web pages in ASP.NET had to be developed. Using these, it is possible for several people to modify the database from a distance, without concern for the synchronization of different copies. The only requirement is access to an internet connection, which is a minor factor today.
Finally, for the public database, we chose to transform the MS Access database into a more robust infrastructure in order to support multiple connections using Microsoft SQL, still with an ASP.NET web layer.4 Every night, the MS Access database (the version that is modifiable online) is first backed up and then duplicated in MS SQL, while the previous version of the public database (accessible only online) is overwritten.
Regarding the statistical exploitation of the database, given that we were working in the Microsoft environment, we chose, for factor analyses, a payable plug-in for Excel, XL-Stat, which offers an interface that is relatively clear and accessible. For the social networks structural analysis, we used Ucinet, the standard software program in 2005, and its graphical counterpart, Netdraw. Looking back on these software programs from a 2017 perspective, now using Gephi, for example, it is clear how much things have moved on.
Could you offer one or two examples of scientific (whether consensual or surprising) results obtained with the help of the database?
One of the major assets of the CIEL databases is the demonstration of the importance of “literary life animators” (see further down) in Francophone Belgium. The project leaders’ intuition at the start tended towards a “weak literary institution” (Aron & Denis 2006); for this reason, they gave a lot of importance, in the structure of the Belgian Francophone literary sub-field, to interpersonal relationships: the systematic collection and serialization of the data helped confirm and elucidate this intuition (Dozo 2011).
With the help of a reconstitution (construction) of the network of authors’ individual relationships, we pinpointed interesting cases using indicators of centrality for each individual. Certain authors have very high indicators of centrality which help distinguish them from others. Among the authors who stand out, some were expected given the previous knowledge of their role within literary history (Franz Hellens, Albert Mockel, etc.). Others, such as Gaston Pulings, Pierre Fontaine, and Paul Werrie were much less so. We named these writers with large tallies of relationships, or with a significant structural relational position (see the concept of structural gaps), “literary life animators” (Dozo 2014).
Another use of the database was the statistical portrait of the social and literary space in Francophone Belgium. By way of factor analyses, we were able to situate active literary generations and their significant characteristics (Dozo 2009), or focus on one profession and study its links to literary practice (Dozo 2010). We could also mention the geolocalization of authors’ residences in Brussels during different periodes of the city’s development (Debroux, Dozo, Vanderpelen 2015).
A public version of the database is available at ciel.philo.ulg.ac.be. Subscription is free of charge and offers the possibility to search by “author,” “work,” or “journal.” The “journals” section offers many digitized resources – the fruits of the second project, ARC CIEL, supported by the Université libre of Brussels under the direction of Paul Aron. Some cross-referenced inquiries are also available (events over one year, places, etc.).