Daejeon, July 27–31
As an interdisciplinary field combining computational methods with humanistic inquiry, Digital Humanities (DH) has developed unevenly across regions, shaped by linguistic, institutional, and technological conditions. This unevenness raises a persistent question: Is DH evolving as a unified global discipline, or does it remain fragmented along regional lines? While the rhetoric of DH often describe it as a “big tent” to emphasis its openness and inclusivity (Svensson 2012), scholars such as Fiormonte (2022) and Risam (2018) have argued that the field remains structurally Anglo-centric, particularly in its publication venues and methodological norms.
Existing attempts to map DH’s intellectual structure often rely on anecdotal observation or focus primarily on English-language scholarship. As Poole (2017) notes, the boundaries and priorities of DH require more empirical investigation. Topic modelling offers a way to statistically examine large bodies of text, making it possible to compare research emphases across regions at scale. Academic journals, as a primary site of scholarly communication, provide a suitable corpus for such analysis.
This short paper presents an ongoing quantitative comparison of DH research themes in the UK and China from 2020 to 2024. Rather than attempting to define DH as a whole, the study focuses on how thematic priorities differ and overlap across national and linguistic contexts, contributing empirical evidence to debates on globalization and regionalization in DH.
The dataset consists of 961 abstracts published between 2020 and 2024 in four DH journals: Digital Scholarship in the Humanities (DSH) and International Journal of Humanities and Arts Computing (IJHAC) from the UK, and 数字人文研究 (Digital Humanities Research) and 数字人文 (Digital Humanities in China) from China. The time frame was selected to ensure comparability, as both Chinese journals were launched in 2020. The sampling of journals is based on the list of DH journals identified by Spinaci and his colleagues, focusing specifically on journals dedicated to DH research (Spinaci / Colavizza / Peroni 2019). Based on this, journals were further filtered by their country of publication and impact factor.
Abstracts were retrieved from Web of Science (WoS) and CNKI. While WoS was utilized for English data in this study, future iterations will explore open-access databases (e.g., OpenAlex) to better align with the FAIR principles. The dataset contains 620 English abstracts (Total words: 639,986; Average: ~1,032 words/abstract) and 341 Chinese abstracts (Total characters: 92,517; Average: ~271 characters/abstract)
Two independent LDA models were trained on the bilingual corpora. The optimal number of topics (k=10) was determined by evaluating the Cv topic coherence score, which peaked and stabilized at k=10. Topics were manually aligned across languages by evaluating the top 30 terms per topic using a relevance score (λ= 0.6). This specific weighting balances overall term frequency with topic-exclusive probabilities, explicitly differentiating adjacent fields. For instance, "Semantic Analysis" and "Literary Analysis" were delineated based on their distinct underlying distributions (metaphor/word vs. narrative/edition), while "DH Infrastructure" was separated from "Computational Tools" by isolating conceptual terminology (theory/domain) from applied methodology (model/analysis).
TPI was applied using step-wise chi-square tests to quantify thematic disparities across the English and Chinese contexts. The initial chi-square statistic (48.53, p = 2.03 * 10-7, d f = 9) rejected the null hypothesis of independence, confirming significant thematic divergence. Iterative exclusion categorized topics into "divisive" (highly polarized) or "bridging" (shared scholarly focus) based on their chi-square contributions relative to the mean. Finally, CA was utilized to project the multidimensional frequency matrix into a two-dimensional visualization. The first two principal components account for 74.38% and 25.62% of the variation, respectively, effectively preserving 99.99% of the total structural information.
The analysis identifies ten core DH research topics across both contexts: DH Infrastructure, Translation Studies, Semantic Analysis, Museums and Arts, Spatial Analysis, History, Cultural Heritage, Literary Analysis, Computational Tools, and Network Analysis.
TPI results show that most topics display uneven distribution between UK and Chinese journals. Chinese journals place greater emphasis on DH Infrastructure and Translation Studies, reflecting priorities in language resources, platform construction, and methodological foundations. UK journals, by contrast, show stronger focus on Semantic Analysis, Museums and Arts, and Spatial Analysis, aligning with established strengths in computational linguistics, digital heritage, and spatial humanities.
Figure 2 Distribution of Merged Topics by Countries
At the same time, several topics (Literary Analysis, Computational Tools, and Network Analysis) exhibit relatively balanced distributions and function as bridging topics across regions. These shared areas suggest a degree of methodological convergence, likely driven by globally circulating tools and standards.
Figure3: Correspondence Analysis of Topics by journals
Correspondence Analysis visually reinforces these patterns. UK journals cluster around analytically intensive and application-oriented topics, while Chinese journals align more closely with infrastructural and language-focused research. Bridging topics occupy intermediate positions, indicating their role in connecting regional DH practices.
These findings complicate binary narratives of DH as either fully global or fundamentally fragmented. Instead, they point to a hybrid structure in which shared technical logics coexist with region-specific priorities. The prominence of infrastructural and translation-oriented research in Chinese journals reflects not marginality but strategic engagement with DH under local linguistic and institutional conditions. This supports arguments that Global South DH practices are not merely derivative but actively shape the field through localized innovation (Chen 2018; Wang / Tan / Li 2020). This empirical evidence partly mitigates the criticism of DH as a “monocultural” field (Fiormonte 2015) and addresses the challenge posed by Earhart (2012) of fostering global collaboration without reducing diverse regional practices to a singular standardized model.
At the same time, the presence of bridging topics suggests that globalization in DH operates primarily through technical and methodological standardization, rather than thematic uniformity. Quantitative methods such as topic modelling thus offer a useful lens for examining how global coherence and regional differentiation are negotiated in practice.
This short paper presents an empirical, data-driven comparison of DH research themes in the UK and China, demonstrating both convergence and divergence within the field. While the study does not claim to represent DH globally, it offers a replicable framework for North–South comparison and highlights the importance of including non-English journals in DH meta-analysis.
Future work will extend this analysis qualitatively, examining how institutional contexts, funding structures, and linguistic infrastructures shape thematic choices. By combining quantitative mapping with critical interpretation, this project aims to contribute to more inclusive and reflexive understandings of global DH.