State-sponsored translation programs play a central yet comparatively underexamined role in shaping global literary circulation. Existing research on translation flows has largely focused on smaller source literatures and on the interplay between target-driven and source-driven dynamics (Levitt and Shim 2022; Vimr 2019; Heilbron and Sapiro 2018), leaving the large-scale, data-traceable effects of centrally coordinated cultural policies less explored. This paper addresses this gap by applying computational methods from bibliographic data science (Gooding et al. 2025; Lahti et al. 2019) to examine how state-driven agendas materialise in the international dissemination of Chinese literature through the Silk Road Literary Translation Program (丝路书香出版工程), a flagship initiative within China’s broader Belt and Road strategy.

Translation has long functioned as an instrument of cultural diplomacy in the People’s Republic of China. While earlier efforts prioritised ideological alignment (Jiang and Ma 2022), contemporary programs emphasise thematic plurality, cultural exchange, and narrative projection under the slogan of “telling China’s story well”. (Jiang 2021) The Silk Road Literary Translation Program, administered by the National Press and Publication Administration, exemplifies this shift while remaining tightly embedded in geopolitical and policy frameworks. It funds translations of Chinese literary works (primarily for export to Belt and Road countries) while supporting overseas translators, sinologists, and publishers.

Situated at the intersection of research on supply-driven translation and cultural-policy approaches to soft power, this paper builds on recent work that conceptualises translation subsidies as instruments of symbolic power (Roig-Sanz et al. 2025). It addresses three interrelated questions: (1) how do state-sponsored translation programs structure global translation flows in terms of target languages, regions, and literary genres; (2) to what extent do these patterns exhibit thematic and cultural diversity; and (3) what can absences, omissions and areas of limited variation (cf. Teichmann and Roman 2024; Wakabayashi 2019) reveal about the selectivity and limits of cultural policy.

Empirically, the study draws on a dataset of nearly 3,000 translation projects approved between 2015 and 2024, published annually in the List of Approved Projects for the Silk Road Literary Translation Program (丝路书香工程立项项目公示名单) in the China Press Publication Radio Film and Television Journal (中国新闻出版广电报). While the original lists provide only minimal metadata – translation publisher, book title, and target language – the dataset was substantially enriched to support multidimensional analysis. Additional metadata include author demographics and professional profiles, publication histories, publisher hierarchies (national versus provincial), and subject classifications derived from Chinese Library Classification codes and National Library of China keyword authorities.

Methodologically, the study adopts a bibliographic data science approach, treating enriched bibliographic records as a primary empirical object rather than ancillary description. Computational analyses combine language and region clustering, genre and subject profiling, and diversity measures such as topic entropy to trace how literary content is distributed across different target languages and geopolitical spaces. Crucially, the analysis integrates the detection of dominant patterns with a systematic examination of blind spots - genres, languages, or regions that are marginal, absent, or unusually homogeneous despite the program’s stated emphasis on diversity.

The results show that while the Silk Road program as a whole supports a broad range of literary types, this diversity is unevenly distributed (see Chart 1). Certain target languages (most notably Arabic) exhibit high thematic diversity, whereas others display sharply constrained topic profiles, suggesting differentiated cultural strategies for different regions (see Chart 2). Moreover, aggregation at the country or language-group level reveals a far more homogeneous selection than the program’s overall rhetoric of pluralism would imply. These blind spots point to the limits of soft-power ambitions and highlight how non-translation (Duarte 2000) functions as a meaningful policy signal.

By combining large-scale bibliographic enrichment with pattern and absence detection, the paper demonstrates how DH methods can offer a nuanced, data-driven account of state-led cultural circulation. Beyond the Chinese case, the study contributes a transferable methodological framework for analysing translation, cultural policy, and soft power as measurable phenomena embedded in bibliographic infrastructures.

Figure 1: Topic proportions by target language including absolute translation count

Figure 2: UMAP of topic profiles by target language