Daejeon, July 27–31
The advent of printing in the 15th century has significantly shaped European history. In the Early Modern period Europe was made up of a number of smaller and larger competing polities, connected by trade networks, religious and dynastic ties, guilds and scholarly communities. The patterns of flow of information between different communities led to globally unprecedented innovation and prosperity in the area (Henrich 2020). The printed word was an important catalyst here: the accumulation and wide diffusion of knowledge led to the industrial revolution (Mokyr 2016), while the formation of imagined vernacular reading communities led to the formation of modern nation states (Anderson 1983). How exactly did knowledge flow between major centers of power in Europe, and how exactly did the national written language communities form?
The abundance of digital data allows these questions to be tackled to the level of detail not possible in traditional scholarship. We use bibliographic data — records of published books — to track both the diffusion of information at the scale of the continent and transitions within communities with great precision. We harness theoretical insights from the field of cultural evolution that seeks to understand cultural patterns through adopting principles and methods from ecology and evolutionary biology (Mesoudi 2016). Darwinian principles of variation, inheritance, and selection offer a powerful paradigm through which to understand the patterns of information diffusion both within a community and over generations. Research in cultural evolution has shown how these ideas can fit well with questions in the humanities and cultural history, while keeping the differences in mind (Sobchuk 2023).
First, to answer the question on knowledge flows, we study the publications of authors across Europe. Each author’s publication history can be seen as an information cascade lasting up to a few centuries. We assess the time it takes for an author to be printed in different regions — proxying the speed at which knowledge spread — and build networks of European cities based on these cascades — tracking the geographic paths of knowledge transmission. To answer the question on new language communities, we investigate the demographics of writer communities when they transition from one dominant language to another (for example Latin to German or German to Estonian). Building on evolutionary biology’s Price equation (which mathematically describes how a trait’s frequency changes over time), we decompose the trends in language use into demographic components in authors’ careers to assess whether shifts occur via the emergence of new authors or changing habits of existing authors. Both analyses adopt established methods from ecology and evolutionary biology to describe trends in cultural history in unprecedented detail, highlighting a promising avenue for future research in the Digital Humanities.
We build our analysis on the Heritage of Printed Book (HPB) database. HPB is an aggregate database based on catalogues of more than 60 libraries maintained by the Consortium of European Research Libraries (CERL). It contains entries on ca. 10 million editions dating from 1450–1830, from the first printed books up to the era of machine press printing. Bibliographic data science (Lahti et al. 2019) has paved the way for systematic analysis of these records. As bibliographic data aims for precision, the records are not standardized for different spellings or forms (Umerle et al. 2022). To analyse the data, we further harmonized the data on author names and place names from CERL and enriched them with updated geographic coordinates. To include data on language communities emerging in the 19th century, we supplement the data with several national bibliographies (e.g. Estonian National bibliography, Kruusmaa et al. 2025).
We track the patterns in publication of authors in different cities to understand the flow of knowledge across Europe. Each author is first published in one or several cities, and if they become known they may then be published in more cities. For example, the Danish-Norwegian writer Ludvig Holberg was first published in Copenhagen (1711) (Figure 1). He was then published in Rostock (1719), Amsterdam (1721), Antwerpen (1723), followed by Cologne (1728), Stockholm (1735), Leipzig (1736), and Heidelberg (1736). The latest cities in the database are Nürnberg (1824), Uppsala (1825), Strängnäs (1828), more than 50 years after the author died. We focus on the most popular authors that were born after 1450 (authors with more than 50 editions, n = 16,574) and track the publication of their works through the recorded editions (n = 3,691,600).
We analyse the data in two ways. First, we look at the distance covered over certain time intervals, building on the “wave of advance” model in cultural evolution used to analyse the spread of major technologies like farming (Ammerman & Cavalli-Sforza 1979). For a particular author like Holberg, we can look at how many kilometers they covered over the years in their career. Second, we look at the typical connections between cities based on social network models used in cultural evolution — specifically network-based diffusion analysis (Franz & Nunn 2009). We build a custom Bayesian workflow that infers a network based on the information cascades of each author (Gomez-Rodriguez et al. 2010).
For both approaches, we group the authors by the starting year into 20-year slices and rely on the first 50 years of their career to make the time periods comparable. Holberg would be grouped with other authors who started 1700–1719, and we only consider the publications up to the year 1761 for him. Figure 2 shows a reconstructed network for authors with the strongest connections visible. The analysis shows strong connections between major publishing centers (like London, Paris, Amsterdam, Rome) and regional attractors (for example authors first published in Dresden, Rostock, and Heidelberg had to go through Leipzig to reach wider fame). We analyse the dependence of these connections on the starting place and time.
We study the demographics of linguistic community formation with a variant of the Price equation (Beheim & Baldini 2012). The Price equation has been used in evolutionary biology to describe the changes in trait frequency in terms of the mechanisms that explain it. For cultural evolutionary data, it has been adapted to investigate how generational and individual change can explain trends observed in literary history (Sobchuk & Beheim 2025). With this method, it is possible to decompose trends to the factors responsible for the changes. We adopt this method to look for the role played by emerging authors, exiting authors, and individual change in changes in language use.
Figure 3 shows the results of the decomposition analysis for two written language communities that emerged at different times, in different places: 1) German from a Latin language background in Germany, and 2) Estonian from a German language background in Estonia. The author demographics show quite different patterns. In Germany, the first leap in German usage documented during the Lutheran reformation (1520–1530) was due to both new authors and individual change, while the second more gradual growth (1730–1780) is largely explained by individual change. On the other hand, the major growths in Estonian language use (1720–1755 and 1830-1900) were largely shaped by new entrants to the written language community. The German community reflects transitions similar to administrative reform, while the Estonian community emerges from an influx of new authors, highlighting the importance of context specific dynamics. We compare the demographics in major shifts in language use across Europe to describe the different types of changes that led to new written language communities in the Early Modern and National Romantic periods.
We present two novel ways that quantitative analyses of bibliographic data can enhance the level of detail in historical reconstruction. The publication histories of authors across Europe can be used to study the knowledge flows between major centers of power. The population of authors represented in language shifts in particular communities can be used to understand the demographic processes that guided these shifts. Many questions in the humanities, while vastly different in topic, are structurally similar to those in the natural sciences allowing for possibly fruitful collaboration. Our methods originate from models in ecology and evolutionary biology that have been adapted for the study of cultural patterns. We hope to demonstrate how cultural historical research can benefit from the integration of such methods.