Daejeon, July 27–31
This paper presents a feminist digital humanities approach to social network analysis (SNA) as a means of recovering forms of cultural influence and creative leadership that have been systematically undervalued within music historiography. Using Marion Scott (1877-1953) – violinist, critic, musicologist, editor, and co-founder of the Society of Women Musicians (SWM) – as a case study, the paper addresses a central methodological challenge in computational humanities research: how to analyse historically marginalised actors whose archival presence is fragmented, dispersed, and shaped by gendered systems of valuation. In doing so, it contributes directly to the conference theme by examining how meaningful connections – between people, institutions, archives, and computational methods – can be made visible through critically engaged data practices.
Marion Scott exemplifies persistent challenges in music historiography, particularly the systemic undervaluation of women’s cultural labour despite archival evidence of their professional contributions (Adams 2011). While Scott achieved public recognition during her lifetime, her historical reputation has often been framed in reductive terms – evaluated through proximity to male composers or dismissed as amateurism – rather than recognised as a form of creative leadership embedded in networks of collaboration, advocacy, and institutional work. Recent scholarship has shown that women in early twentieth-century musical culture frequently attained professional authority while simultaneously encountering formal exclusion from male-dominated institutions, complicating assumptions about musical modernity and professionalisation (Fuller 2018).
These historiographical patterns raise deeper epistemological questions central to feminist digital humanities and feminist data science: whose knowledge is preserved, whose networks are rendered legible, and how data collection and modelling practices can reproduce or challenge existing power structures (D’Ignazio / Klein 2020). From this perspective, engagement is not merely a matter of scholarly outreach or participation, but a critical stance towards data itself as a historical and social artefact. This paper argues that SNA, when grounded in small-data methodologies and interpretative annotation, offers a way to re-engage with archives in ways that resist overgeneralisation.
The paper is situated within the broader “network turn” in humanities scholarship, which has reconceptualised cultural production as fundamentally relational (Ahnert et al. 2020). This shift aligns with current research that moves beyond score-based analysis towards understanding music as performance, practice, and social interaction (Cook 2013). Musical culture emerges through dense webs of collaboration, mentorship, editorial labour, institutional affiliation, and informal networking – forms of engagement that are rarely captured by traditional metrics of achievement (Crossley et al. 2014). Recent database-driven studies have demonstrated the value of network approaches for uncovering such dynamics, from Willen’s mapping of Stockholm’s musical life (Willen 2024) to Koivisto’s large-scale analysis of European ladies’ salon orchestras, which illuminates intersections of gender and class that resist canonical narratives (Koivisto 2019).
Despite these developments, musicology has been comparatively slow to adopt computational methods, particularly in ways that centre women’s professional networks rather than canonical works or institutions. Automated techniques for extracting relational data from unstructured texts have proven effective in other humanities domains (Graham et al. 2020), yet historical musical sources often require careful manual interpretation due to the complexity of professional relationships and the specialist knowledge needed to contextualise musical practice. This paper responds to this gap by presenting a manually curated, small-data network as both a substantive contribution to early twentieth-century music history and a methodological intervention in how women’s cultural influence can be studied computationally.
The empirical foundation of the study is a purpose-built dataset derived from extensive archival research at the Royal College of Music. The project undertook a comprehensive survey of three major archival collections in order to capture Scott’s multifaceted role as a network broker within London’s musical community: (1) the Marion Scott Collection (87 catalogue entries), comprising personal correspondence and papers; (2) the Society of Women Musicians archive (526 catalogue entries), documenting the organisation Scott co-founded and helped to run; and (3) the Royal College of Music Magazine archive (264 entries from issues edited by Scott). From this survey, 76 documents were strategically selected based on their capacity to evidence Scott’s role in connecting disparate musical communities, including marginalised composers, women performers, institutional gatekeepers, and mainstream publications. This selection process adapts established principles of historical data extraction to the specific requirements of network analysis (Whitfield 2023).
Rather than focusing solely on Scott as an individual subject, the project systematically documented all identifiable relationships referenced within these materials. Following recent scholarship that emphasises the analytical value of examining what is hidden, excluded, or marginalised within archives (Barget / Schreibman 2025), each document was examined for evidence of professional, social, and collaborative ties. Particular attention was paid to relationships that challenge canonical assumptions, including editorial advice, concert organisation, committee work, mentorship and advocacy, social gatherings with professional implications, collaborative performances, correspondence concerning professional matters, and shared institutional affiliations. This emphasis responds to growing interest in women’s collaborative and supportive practices within musical cultures, areas that traditional scholarship has only recently begun to address in sustained ways (Mathias 2021; Hamer 2021; Head / Wollenberg 2024).
The resulting open-access dataset comprises 504 individuals connected by 618 documented relationships spanning the period from 1900 to 1959. Although Scott served as the initial entry point, the network expanded organically as connections between third parties were recorded, allowing the analysis to move beyond an individual-centred perspective.
Each identified relationship was recorded with detailed metadata, including the names and unique identifiers of connected individuals, the nature and directionality of the relationship, relevant dates, documentary evidence type, archival reference, geographic location, event context, and interaction direction. A standardised weighting scale from 1 to 5 was developed to capture relationship intensity: 1 denoting figures referenced without evidence of direct interaction; 2 indicating professional acquaintances or single documented interactions; 3 encompassing teachers, mentors, and regular professional contacts; 4 representing sustained peer-level engagement; and 5 identifying intimate professional partnerships such as long-term collaborators or co-founders. This classification system enables both qualitative interpretation and quantitative analysis while preserving transparency regarding interpretative decisions.
The paper employs three complementary approaches: centrality analysis, community detection, and temporal network analysis. Degree and betweenness centrality are used to identify structurally significant actors and intermediaries within the network. Community detection is conducted using the Louvain algorithm, revealing clusters of densely connected individuals corresponding to different professional and institutional groupings. Temporal segmentation allows the network to be analysed across decades, enabling correlation between structural change and historical events.
The network exhibits low overall density (0.0024), indicating a sparse structure in which connectivity relies heavily on intermediaries. Marion Scott emerges as a highly connected node, with a degree centrality of 0.6243, confirming her role as a crucial connector. More striking, however, is the gendered distribution of centrality: four of the five most central figures in the network are women – Marion Scott, Katherine Eggar, Fanny Davies, and Liza Lehmann. The prominence of Fanny Davies is particularly revealing. Known primarily as a concert pianist, Davies emerges as the third most central figure, functioning as a bridge between pedagogical institutions, performance networks, and advocacy circles. Her centrality was not anticipated and was not the result of targeted data collection, illustrating the capacity of SNA to surface overlooked forms of connective labour.
Community detection further challenges narratives of women’s musical activity as marginal or isolated. Eleven distinct communities are identified, each representing different loci of musical practice. Crucially, the most central women are not confined to single clusters; instead, they bridge multiple communities, facilitating collaboration and information flow across otherwise weakly connected groups. This pattern underscores the importance of relational and infrastructural labour – forms of work that are essential to cultural production but rarely foregrounded in traditional historiography.
Temporal analysis reveals significant structural change during the 1910s, with a marked increase in network density and interconnection coinciding with the founding of the Society of Women Musicians in 1911. This shift suggests that the SWM functioned not merely as a symbolic organisation but as an active alternative professional network to male-dominated institutions, enabling women musicians to circumvent formal barriers through collective infrastructures.
The paper concludes by arguing that feminist SNA offers a productive framework for rethinking cultural influence as relational, collaborative, and historically situated. By making visible the connective labour performed by women such as Scott, Eggar, Davies, and Lehmann, this approach challenges individualist and output-centric models of cultural value and demonstrates how critically applied computational methods can reshape our understanding of musical culture and women’s creative leadership in the early twentieth century.