The modern notion of early modern English dramatic genres has been heavily shaped by canonized authors and texts. Yet approaches grounded in book history increasingly reveal that a wide array of marginalized agents, such as stationers (printers, publishers, booksellers) and non-canonical playwrights, also played substantial roles in shaping a play’s genre across its textual life cycle. One important method of recovering the genre-shaping activities of these agents is to read the paratextual elements of early playbooks. We use “paratext” here to refer to the liminal devices and coventions (Genette 1997) of the playbook that frame and guide the reception of the playtext, including play titles, title-page attributions, and stationers’ information.

While recent literary scholarship has emphasized close readings of individual paratexts and their relation to genre (Kesson 2014; Gilchrist 2021; Lidster 2022), computational approaches offer a complementary large-scale perspective capable of creating new frameworks for understanding these materials. Aside from computational analyses of genre focusing on play texts, such as natural-language-processing-driven linguistic analysis (Hope & Witmore 2010), function-word network analysis (Eisen et al. 2018), quantified character-data modeling (Amelang 2025), and synthesized computational-philological analyses (Lamb 2020), large-scale studies of paratextual elements of playbooks have also gained momentum in recent scholarship as aggregated data become more accessible. The open-source database DEEP (Database of Early English Playbooks) has been especially valuable for such work because it provides multi-layered genre classifications (Farmer & Lesser 2008), which are partially incorporated or adopted in recent digital humanities research such as Smith (2023) and Dahl (2023).

Building on these explorations of DEEP (Farmer & Lesser 2024), we developed this project to further investigate the dataset’s potential of visualizing larger-scale and networked paratextual data. Through statistical inference and social network analysis of the full set of DEEP’s 1,921 records of early modern English drama, we seek to complement micro-level paratextual studies with a macro-level account of how aggregated paratextual information illuminates the complex social strata underlying the genre identities encoded in early modern English playbooks.

We first extracted genre information for all records in DEEP and manually classified these genres into eight major categories, drawing primarily on title-page genre designations and consulting the major modern genre compendium used by DEEP when no title-page genre was available. We then extracted the first publication date for each record and grouped these dates by decade, combining all decades prior to the 1560s into a single group and all decades after the 1650s into another. Next, we visualized the temporal variation in genre proportions based on the first publication date of each record.

Figure 1. Temporal Variation in Genre Proportions by First Publication Decade

The overall trends shown in Figure 1 confirm several well-established patterns in literary history, such as the popularity of comedy in the 1600s (Gurr 1980) and the prominence of history plays in the 1590s before their sharp decline after 1610 (Griffin 2001).

However, large-scale paratextual analysis can reveal the latent social strata of genres, which may or may not corroborate commonplace narratives of literary history. To begin demonstrating this, we extracted author and stationer information for each record and applied Fisher’s exact test (Fisher 1922) to identify author–genre and stationer–genre combinations whose frequencies depart significantly from random expectation: that is, cases where an author writes or a stationer produces a particular genre far more frequently than would be expected by chance.

Our results on author–genre specialization both corroborate and complicate established literary-historical narratives. The top three author–genre combinations, ranked by statistical significance, are Ben Jonson–Entertainment, William Shakespeare–History, and Richard Brome–Comedy, all of which align with long-recognized accounts of these authors’ characteristic styles. However, the analysis also reveals cases in which authorial specialization appears less stable than the scholarly tradition suggests. For instance, Thomas Middleton, often reputed as an outstanding tragedian, emerges in our results as having a more contingent genre profile than previously assumed: Middleton–Tragedy ranks 528th among the 537 author–genre combinations. While tragedy constitutes 25% of all author–genre combinations overall, it makes up only 14% of Middleton’s. Marked by extensive collaboration and a largely freelance mode of authorship, his career illustrates how genre specialization, and even authorial identity itself, can be shaped by institutional and commercial conditions rather than by individual choice alone.

Similarly, our analysis of stationer–genre specialization highlights previously underexamined patterns in the role of publishers, printers, and booksellers in genre formation. Consistent with what recent literary studies have shown (Lidster 2022: 132–179), our results identify Nathaniel Butter–History as the most statistically significant stationer–genre combination (p=1.2×10⁻¹²). By contrast, the second most significant combination we discovered—Thomas Purfoot (Junior)–History (p=3.3×10⁻⁸)—is markedly underrepresented in scholarship. While previous studies have focused on Purfoot’s father, Thomas Purfoot (Senior), and his involvement in printing politically topical and historical works (Serjeantson & Woolford 2009), little has been said about how Purfoot (Junior) may have continued this family tradition—precisely the kind of avenue our results suggest is worth exploring. Notably, the fact that the top two stationer–genre combinations ranked by statistical significance both involve the genre of History also illuminates history plays’ characteristics as a distinct dramatic category with a coherent print identity, shaped by agents with unusually strong specialization in publishing this genre.

Subsequently, we analyzed the social network of stationers using paratextual data to further investigate how connections like Butter–History and Purfoot–History were socially situated. We first connected stationers to authors with whom they collaborated on at least one drama and then projected this bipartite network into an inter-stationer network, where nodes represent stationers and edges represent cases in which two stationers collaborated with at least one common author. The two stationers discussed above, Butter and Purfoot (Junior), are found to be closely connected: they share collaborations with five common authors. Among all 199 stationers connected to Purfoot, only two share a higher number of common authors; and among the 187 connected to Butter, none share more. This suggests that Purfoot’s pronounced specialization in history may not stem solely from family inheritance, but also from his embeddedness within a history-play-oriented print/publisher community. Such findings, unlikely to emerge from close readings of individual playbook paratexts alone, invite further archival investigation into trade connections similar to that between Purfoot and Butter.

Figure 2. Stationers’ Network by Dominant Genre (Node Color) and Community (Node Shape)

In Figure 2, we visualize the stationers’ network, where node colors indicate the genre each stationer most frequently published and node shapes mark communities detected using the Louvain algorithm (Blondel et al. 2008). Of the 405 stationers, 383 (95%) form a large interconnected component, while the remaining 22 appear as marginal islands that are unreachable from the main component (shown fully at https://anonymous.4open.science/r/Early-Modern-English-Drama-25D4, where zooming out reveals their placement). The visualized network invites attention to these marginalized figures in the playbook trade. For example, Thomas Vautrollier, the Huguenot exile–printer appears as an isolated node with no shared-author connections to other stationers, suggesting a sparse degree of connection between English stationers and the transnational playbook trade.

More importantly, the community structure shows how clusters of stationers jointly stabilize the print identities of specific dramatic genres. At first glance, Figure 2 suggests little correspondence between node colors and node shapes: stationers who most frequently published the same genre do not necessarily form tightly connected communities based on shared-author collaborations. This impression is supported by the normalized mutual information score of 0.117, which indicates a very weak relationship between the two partitions. However, the Kulczynski similarity scores between individual stationers’ dominant genre and network-community pairs reveal a notable pattern. The three highest similarities involving a central community (defined here as communities within the main component shown in Figure 2) are all associated with stationers who most frequently published comedies. While comedy is often described as an unstable or constructed genre in micro-level analyses of playbook paratexts (Kesson 2014), our results indicate that it was also a socially reinforced and community-responsive genre with a clear network identity, supported not only by individual book-historical artifacts but also by stable, multi-agent infrastructures of playbook production and circulation.

In conclusion, complementary to the approach of viewing early modern English dramatic genres as context-dependent categorizations that vary from one playbook paratexts to another, our method allows us to uncover the more durable social strata in the genre-shaping processes. Using DEEP’s aggregated paratextual data, our network and community analyses help recover the collaborations and activities of heterogeneous and previously overlooked stationers, bringing these figures from the margins of playbooks back into the center of the social processes of genre formation. At the same time, we recognize that our approach is shaped by the inherent limitations of playbook-based data, and in future work we plan to incorporate performance record data, such as REED (https://ereed.org/), in order to investigate performance-based genre identities alongside print playbook data.

Nevertheless, we believe that a dedicated aggregated paratextual resource such as DEEP, combined with statistical inference and large-scale social network analyses, offers a powerful framework for understanding the print genre identities of early modern English drama—identities shaped not only by the Shakespearean canon, but also by diverse non-canonical works and the stationer communities that supported them.