The SIG aims to sponsor a new type of digital/computer-aided analysis of literary style. To do so, it facilitates inter- and transdisciplinary interaction between different fields of “style studies,” including computational stylistics, authorship attribution, corpus stylistics, and digital hermeneutics. Thereby it fosters novel empirical findings, methodological and epistemological innovation, as well as advocacy and networking. It is interdisciplinary by nature.
The main goals of the SIG-DLS are:
- To define important, interesting, difficult questions for our group
- Discussion of values and priorities (i.e., ‘what makes a digital literary scholar?’)
- Outline a set of best practices for our research
- Actively liaise with other, relevant initiatives
- Provide an online home/hub for information about:
- Best practices
- Bibliographical information
- Repository of seminal papers
- Events, including Challenges
- Research of interest
- Sharing corpora and code
- Grants and Jobs
The SIG will take a strong stance on multilinguality, promoting and encouraging the cooperative creation of corpora for shared tasks in different languages.
Motivation (from original proposal)
For centuries, the study of the style of literary works has been a core activity of literary critics and philologists across different scholarly communities. The study of style — simply understood as “manner of writing/speaking” — is both an important descriptive category for humanist scholars and a productive instance of meaning-making in culture.
There exists a considerable tradition of the study of textual style by computational means, especially if we consider “computational” in its etymological sense of ‘counting’, not necessarily aided by a machine. In particular, forensic linguistics and other fields have long established means for measuring textual patterns in order to answer classification questions. Despite this historical rooting of “computational stylistics”, such approaches have recently been able to leave the sphere of forensics to become an instrument of legitimate literary analysis of style.
One of the oldest branches of computational humanistic studies, Digital Literary Stylistics is now an area of expanding inquiry and it offers a way to bring together exciting and diverse work in computational stylistics, authorship attribution, corpus stylistics, NLP, and digital hermeneutics.
In view of the growing interest for digital methods in the study of style, we propose to the Alliance of Digital Humanities Organizations (ADHO) to set up a SIG on “Digital Literary Stylistics”, in order to bring together researchers from different perspectives to discuss theoretical, methodological and technical issues; share resources; and organize events and initiatives.
Tapping into these traditions, the proposed SIG will shed a fresh look at style. Literary style, as a pivotal cultural phenomenon, is a key descriptive category of academic study that now may be analyzed using arithmetic and algorithmic methods. This is not least due to the fact that large archives of digitized literary texts have become freely accessible and digital tools for text analysis are increasingly powerful, yet easy to handle.
While the SIG-DLS will have a clear focus on literary style, it will be a platform fostering the study of style in any type of discourse, register, and genre. This means we decidedly include the study of non-literary texts. From the current point of view, any description and theory of “literary texts”, including any sub-categories and sub-groupings, benefits greatly from comparison to other domains of discourse (e.g., religious, journalistic, juridical, or political). A description of the “literariness” of singular texts, and groups of texts, is validated by a sufficient description of the “non-literary” (and its many facets). This holds true at quantitative and qualitative levels of description.
The need for taking a cross-discourse comparative perspective becomes even more obvious when considering the practices of domain adaption of NLP-tools: We increasingly understand that many literary texts, or possibly individual literary genres, have properties that make it necessary to take their specific makeup into account when designing computational methods for their analysis. For example, the algorithms used for Named Entity Recognition and Coreference Resolution that have been traditionally trained on newspaper materials cannot be used out of the box for novels. However, when running these types of automated analyses, we do not reinvent the wheel by designing entirely new procedures, but act upon methodologies developed for adjacent domains of discourse. Eventually, a reverse direction of tool adapation might be envisioned, with methods originating from algorithmic Digital Literary Stylistics being adapted for usage for other domains of NLP.
The time seems thus ripe to offer a space where the aesthetics- and hermeneutics-oriented approaches to style can interact with the computational ones, zooming in on a wide array of possible patterns of literary language use (and their potential functions in discourse and cognition).
The wide interest in the topic within the DH community is well documented by the number of related papers at ADHO conferences. However, despite the evident importance of literary “style” analysis (within academia and society) and its new potential brought by vast digital data and statistical power, there is a lack of theoretical background (e.g., definition of “style”, “author”, “genre” etc.) and methodological standards. Also, people interested in style and using digital methods often come from slightly different traditions, including some that have not yet established digital commons. These factors have hindered seamless communication and successful collaboration.
The newly created “Digital Literary Stylistics” SIG will aim at bridging this gap, providing anyone interested in digital studies of style with a hub for conducting research that is both methodologically and theoretically sound. It also caters to research situated within an active and innovative peer community. This applies to scholars from “technical” (computational) backgrounds, as well as to those who have only recently formed an interest in the digital study of style.
As such, the SIG sees itself providing a core “DH” service – enabling cutting-edge research at the interface of hermeneutic and “computational” communities. Together, we will develop research infrastructures (text analysis pipelines; encourage use of community hubs such as GitHub, Zotero etc.), build and share corpora, and develop a shared conceptual and technical vocabulary to ensure that scholars taking distinct disciplinary approaches to style can grow from and build on one another’s work.