What is abstraction? The word, which means literally ‘to drag away’, is itself a highly abstract and ambiguous concept, taking on distinct yet crucial roles within fields as diverse as art, linguistics, psychology, finance, critical theory, philosophy, and computer science. Within artificial intelligence, for instance, abstraction is arguably the goal of the entire enterprise, as data scientists train machine learning algorithms to ‘abstract’ robust patterns from their input data in order to generalise to unknown samples. Likewise, research in the digital humanities and ‘distant reading’ is often made possible by a process of abstraction: by designing algorithms to extract data from thousands of texts across centuries of literary history, researchers abstract away what is singular and unique about individual texts in order to isolate the larger-scale historical patterns which they collectively form. In this talk, I present such distant data for the usage of abstract and concrete language across the history of prose fiction. This data measures the extent to which stories are told with abstract words like ‘sincerity’ and ‘resentment’, or with concrete, physical words like ‘piano’, ‘barometer’, and ‘chair’. In turn, this experiment on abstract language lends a ‘meta’ or self-reflexive goal to my project on abstraction: by tracing the history of abstract language in literature, one can uncover a larger history of abstract thinking within which our contemporary culture of abstraction—including even current developments and debates in artificial intelligence—can be better positioned and understood.

Ryan Heuser is Research Fellow in King’s College at the University of Cambridge. His work applies computational methods to the study of literature and its history, with a focus on the long eighteenth century in Britain. He received his PhD in English Literature from Stanford University, where he was also a founding member and long-time contributor to the Stanford Literary Lab.


The Observatoire des textes, des idées et des corpus (ObTIC), SCAI's project-team dedicated to digital humanities, draws on an established expertise in the domain of digital editions, data exploration and production for humanities research. ObTIC team members are actively involved in the design and application of software and algorithms for humanities researchers, as well as the development and evaluation of new digital research methodologies across these same fields of inquiry.

ObTIC’S specificity: opening up traditional humanities disciplines to digital methodologies by approaching texts and corpora through transversal concepts independent of any particular textual genre.