New kinds of sources and tools
This special issue on "Wikipedia, Wikidata, and World Literature" revolves around encyclopedic data and interlinked facts that can provide novel sources and tools for studying the reception of world literature. It brings together five contributions that offer new insights into canons and counter-canons, but also address systematic gaps.
Using Wikipedia and Wikidata as resources allows us to reshape current scholarship on literary canonicity and popularity, which is too often blinkered by abstract notions of influence and implicit bias. Despite the longstanding debate over the canon, what Wikipedia and Wikidata show us is that there is no monolithic canon, but many canons, depending on the data you choose to examine. Is Shakespeare canonical? Yes, unless your corpus is the more than 100 Wikipedia language editions without an entry on Shakespeare.
Since Goethe’s famous words on world literature in his conversation with Eckermann on January 31, 1827, the concepts of “world” and “literature” have been investigated and argued over. The Eurocentrism embedded in comparative and world literature studies – what Franco Moretti famously referred to as work that is “fundamentally limited to Western Europe, and mostly revolving around the river Rhine (German philologists working on French literature)” (Moretti 54) – has given way to broader perspectives and different contexts. Important, in this regard, has been the development of the concept of “worlding” (as summed up in D’haen 12–27), which exposes the putative objectivity of the scholar of world literature.
At the same time, the advent of digital humanities has helped to find new ways of approaching the entire debate, through large-scale literary analyses that tap into online resources like Google Books, the HathiTrust Digital Library and other full-text corpora, as well as reader-oriented platforms like Goodreads, library catalogs and, more recently, Wikipedia, the world’s largest encyclopedia, started in 2001, and Wikidata, the immensely growing knowledge graph, launched in 2012.
The first attempt at a large-scale examination of the reception and representation of world literature through Wikipedia was a 2017 article by Hube et al., a team of scholars (two of whom are among the guest editors of this special issue): the main author, data scientist Christoph Hube, was joined by the computer scientist Robert Jäschke, scholars of digital humanities Frank Fischer and Gerhard Lauer, and world literature scholar Mads Rosendahl Thomsen. It was indicative that this team comprised specialists from different disciplines, all working together towards opening up this research field.
Their approach was to focus on 15 Wikipedia editions, which they comprehensively analyzed using DBpedia, a community project which aims to extract and provide structured information from Wikipedia. Wikidata was still at an early stage at that time and could not have been used to design their study in the same way. Working with DBpedia data from 2014, they ranked the most prominent authors using quantitative and network metrics (page length, number of in-links, PageRank writers, PageRank complete, and page views during 2012, 2013 and 2014). The article paved the way for other articles on the topic as well as this special issue of the Journal of Cultural Analytics.
A simpler approach was used in a 2017 monograph on Finnish writer Aleksis Kivi (using Wikipedia data collected in 2015) by Douglas Robinson. Here, the number of Wikipedia language versions of an entry on an author or literary text was introduced as part of the “Metrics of World Literature” (Robinson 43–83). Since then, counting the number of Wikipedia language versions on a literary text has been established as “a simple measure of canonicity” (Kukkonen).
Around the same time, smaller projects were presented at digital humanities conferences approaching the new possibilities from other angles. In a contribution to the DH2016 conference, Miller et al. set out to identify a subset of significant works of world literature and mine the corresponding Wikipedia articles including their discussion pages to obtain material for comparative studies. Their idea is showcased by a comparison of ten topics extracted via topic modeling from the discussion pages of the English and Italian Wikipedia articles on Homer’s Odyssey.
Another paper presented at the EADH2018 conference concentrates on Dutch literature. Lucas van der Deijl and Roel Smeets collected outlinks from the 2,286 articles in the Dutch Wikipedia at the time that were labeled with the category “Dutch author” (“Nederlandse schrijver”). The outlinks were then used to build a graph to determine which Dutch authors were canonical based on network centrality metrics. They found that “the crowd in many cases merely reproduces the preference for authors that have featured [on] the reading lists of ‘our teaching institutions’ for decades.”
Studies that followed these initial attempts mainly built on the approach of Hube et al. – Paula Wojcik and Sophie Picard’s commentary on this paper provided theoretical reflections on the core literary concepts of “classics” and “canon.” They highlighted the potential of Wikipedia studies to shed light on the habits of Wikipedia authors and non-writing users and in this way to gain a differentiated picture of canon formation.
Along those lines, Mads Rosendahl Thomsen characterized Wikipedia as “a prominent example of a source for peeking into the wisdom of the masses, rather than the preferences of a few,” when describing the processes of canonization within discourses of world literature (Thomsen, “Changing Spaces” 57; see also Thomsen, “Media and Method”).
Meanwhile, in a series of articles, Jacob Blakesley drew from Hube et al. to carry the research forward in a number of different case studies: from the Wikipedia reception of Shakespeare and Dante, to Joyce and modern Italian poets. Unlike Hube et al., Blakesley directly queried 300-some Wikipedia language editions, and aimed to show the lack of global popularity of these authors (Blakesley, “The Global Popularity of William Shakespeare in 303 Wikipedias”; Blakesley, “World Literature According to Wikipedia Popularity and Book Translations”; Blakesley, “The Global Popularity of Dante’s Divina Commedia”; Blakesley, “The Wikipedia Popularity of James Joyce”).
In their forthcoming paper “Circulation and value creation using the example of literary characters”, Sophie Picard, Paula Wojcik, and Sina Zarrieß conducted a preliminary study for the article included in our issue. The authors compared different types of literary value by using literary characters as an example. In particular, they compared a list of characters from a lexicon of literary characters, lists from popular books that rank characters according to their significance in readers’ lives, and the ranking of characters according to Wikipedia. They thereby showed how canonical, popular, and collaborative types of valuation relate to each other.
The increasing importance of knowledge graphs is highlighted in two recent papers. The first one is a review of the role of Wikidata in digital humanities projects. There are three main applications: as a content provider, as a platform, and as a technology stack (Zhao). Out of the 50 projects referenced in the paper, 45 use Wikidata as content provider, much the same as some papers featured in this special issue. This will hopefully encourage literary scholars to discover the possibilities of Wikidata for their research because, as stated in the aforementioned article, “the greater a domain’s usage of Wikidata, the more likely its breadth and depth will increase on Wikidata” (Zhao 20).
The second recent paper we would like to mention introduces the Under-Represented Writers Knowledge Graph (URW-KG), a discovery tool addressing the underrepresentation of non-Western writers. The project seeks to align data from Wikidata with three other resources, namely Goodreads, Google Books, and Open Library (Stranisci et al.).
As we can gather from this brief literature review, there are certainly examples of research using Wikipedia and Wikidata as sources and tools for world literature studies. However, we felt that the potential has not begun to be fully realized. This prompted us to propose this special issue to discover whether there is more interest in this new theoretical paradigm. We were pleased to have a number of scholars from different international backgrounds contribute to our issue, which shows that this paradigm is slowly making its way in the field.
The five articles presented in this special issue share important commonalities, but also point to different ways of analyzing literary reception through Wikipedia, as our brief summaries will show.
Overview of this special issue
Melanie Conroy examines the notorious gender gap in Wikipedia from a new perspective in her article “Quantifying the Gap: The Gender Gap in French Writers’ Wikidata.” In recent years, attention has been drawn to the few female Wikipedia editors (in particular in admin positions) and the harassment they face, as well as to the representation of topics considered typically female and to biographies of women: their number, length, neutrality, quality, and the risk of being deleted. With a particular focus on French women writers, Conroy not only analyzes their representation on Wikipedia, but also sheds light on the “ways in which women are integrated, or not, into Wikidata’s knowledge graph.” Using Wikidata sitelinks and statements, Wikiquote and Wikipedia links, she measures the impact of about 5,000 writers “in ways that contribute to world-historical narratives like national literatures, periodization, and spatial influence.” Conroy’s research shows that the historically given gender gap in printed encyclopedias is echoed in the Wikidata graph. There are only a few writers, like George Sand, who are represented on a scale that can be described as “global.” These articles are written individually and are not merely translated, and the writer’s pseudonyms are linked properly. Most French-language writers are represented in only one or a few language editions. Moreover, Conroy’s data confirms previous research on Wikipedia, which shows that historical topics are underrepresented in the online encyclopedia: only French women writers who were active after 1800 are adequately represented. The article concludes with some pragmatic recommendations on how to close the gap.
Paula Wojcik, Bastian Bunzeck, and Sina Zarrieß, in their study “The Wikipedia Republic of Literary Characters,” explore how the world of literature presents itself when it is viewed through the lens of literary characters, or rather their representation in Wikipedia. By referring to Pascale Casanova’s groundbreaking work La République mondiale des lettres (1999) and working with a network of 7,000 characters featured on more than 19,000 independent character pages, this article demonstrates the transcultural entanglement of languages by means of literary characters. It points to the fact that key concepts of world literature studies in general and Casanova’s model of the world in particular, are in need of adjustment. It also challenges the presumed center–periphery opposition and/or the canonical status of authors, works, and genres by means of an approach that is double-focused on users: through the collaborative platform and through the choice of characters – the literary unit with which readers most frequently identify with. Additionally, this shift in perspective demonstrates the relevance of nationally and transnationally organized fan communities and so-called minor language editions for the representation of literature in Wikipedia.
Matylda Figlerowicz and Lucas Mertehikian, in their article “An Ever-Expanding World Literary Genre: Defining Magic Realism on Wikipedia,” contribute to the discussion of what world literature is or what it means through the lens of magic realism as represented on Wikipedia. Magic realism is a disputed genre which oscillates between narrow (focused on its cultural and geographical origins) and broad (focused on aesthetics) definitions. The authors explore how Wikipedia articles from 56 different languages and cultural traditions represent magic realism as a genre of world literature. Specifically, they perform a close reading of definitions of magic realism in those articles to identify main themes and implicit premises. They also analyze lists of writers and works, their mentions in articles, as well as similarities between articles in terms of the writers they mention. This reveals that “narrow and broad definitions of magic realism compete and overlap in Wikipedia” and that articles draw upon a variety of (scholarly) references, often including (or even focusing on) writers from the articles’ language, even if they are not generally considered representatives of magic realism. The authors argue that the concept “glocal” shows similarities to magic realism in how it is defined and discussed and that it can also be used to frame the circulation of magic realism.
In her paper “Italian Nostalgia: National and Global Identities of the Italian Novel,” Anna Sofia Lippolis sets out to compare the importance of contemporary Italian literature between 1980 and 2021 in local and global contexts. Her starting point is the realization that the nationally declared canon and the global perception of Italian literature appear to be at odds with each other. She applies easy-to-follow metrics like counting language versions of articles, number of in-links and average ratings to measure the popularity of Italian novels on Wikidata, Wikipedia and Goodreads and compares these results to the rankings of traditional national literary awards. User-driven digital platforms function as mediating instances which broaden the debate around the national literary canon and reflect the significance of contemporary Italian literature in a more international framework. The workflow underlying her paper is readily transferable to literature in other languages and countries.
In his article “Escritor / Qillqaq: The Representation of Peruvian Literature in the Spanish and Quechua Wikipedias,” Daniel Carrillo Jara investigates the immense differences in the representation of Peruvian writers in these two Wikipedia editions. He shows that heavily-populated and well-off Peruvian regions are better represented in both Wikipedia editions, namely that socio-economic status influences representation online. But, he also finds that there is no strong correlation between digital accessibility and literary representation. At the same time, he shows that both the Spanish and Quechua Wikipedia editions reflect academic biases, such as the exclusion of Amazonian writers. However, he also discovers that the representation of writers in the Quechua Wikipedia edition is far different from that of the Spanish one, including two dozen Quechua writers absent from the Spanish edition, whose most popular authors are not identical. In short, both Wikipedia editions “propose different narratives of Peruvian literature.” This case study illustrates that the Quechua Wikipedia edition, made up of “the participation of diverse communities,” is busy “constructing a national literary tradition.” While we must not forget the structurally unequal elements undergirding Wikipedia, the online encyclopedia is nonetheless an important means for minority and regional language communities to build up their online presence and shape their literary traditions.
In the following discussion, we highlight four transversal aspects of the world literature debate addressed in the articles.
Literature and the world
The first aspect concerns the bigger picture. There are a variety of (sometimes contradictory) concepts on how to define and structure world literature. Some – like Goethe, who coined the term Weltliteratur – highlight its heterogeneity, while others – like Erich Auerbach and Franco Moretti – rather point to a homogenization of literary standards. Some – like David Damrosch, Venkat Mani, Ankhi Mukherjee, and Sandra Richter – focus on dynamics such as exchange, circulation, dissemination, and/or appropriation, while others – like Harold Bloom and Pascale Casanova – favor rather predetermined concepts which oppose center and periphery or address universal canonicity. The evaluations and thus the answers to questions of whether the respective concepts are future-oriented or reactionary, whether they testify to high moral standards, to a cultural imperialism, or to “epistemic violence” (Spivak 155), whether they depict a desirable or a fundamentally unjust world, whether they connect or separate the world also vary depending on the position. The articles in our issue confirm that the world of literature in Wikipedia and Wikidata does not present a greatly improved version of reality. Accessibility, censorship, economic conditions, etc. determine which literatures are represented and to what extent. According to the status quo of the world literary field (Sapiro 488), African literatures are hardly represented (with the exception of globally acknowledged writers such as Wole Soyinka, Chinua Achebe, Ngũgĩ wa Thiong’o, or Abdulrazak Gurnah). This issue reflects on divisions and inequalities between the global North and South, between economically better and worse developed regions, some of which are reinforced by a digital divide: see, for example, Daniel Carrillo Jara’s article on Peru, which notes a significant digital divide among regions. Among other aspects, this issue also discusses how the marginalization of literatures and literary languages is reflected in Wikipedia and it also addresses a marginalization that, although well-known and mitigated by numerous efforts, persists: the underrepresentation of female writers, as exemplified in Melanie Conroy’s article.
Canonicity, then, is the second aspect repeatedly addressed in our issue. Considering the long history of canon debates and even wars, Wikipedia can provide some relief. As a collaborative and bottom-up-oriented encyclopedia it presents a plurality and diversity of canons, the absence of which is lamented in many analyses of world literature. It unites the perspective of advocates of so-called high-brow culture with that of fans of different genres. The articles we gather in this issue discuss how Wikipedia mirrors the hegemony of a canon of white European males and at the same time how it presents localized world literary canons. They show that Wikipedia challenges the rather monolithic idea of genre in canon debates, and how, for instance, fandom culture works as the “invisible hand” (Winko) in processes of canon formation. But again, we must emphasize that Wikipedia does not radically correct the conditions of the literary world as it persists in the educational and academic system. New entries in Wikipedia must meet the notability requirement, and notability is, at some level, a measurement of value. This requirement limits the amount and breadth of items included in Wikipedia but some younger contributors may also partially define popularity as indicative of value. The cultural capital gathered in both Wikipedia and Wikidata reflects awards and prizes, academic scholarship, translations, textbooks, historical-critical editions, reviews, etc. on the one hand, and media adaptations, blogs, fan cons, fan wikis, commodity culture, etc. on the other. At the same time, an underrepresentation of historical authors and literature can be observed. In her 1985 essay on the question of literary value, Gayatri Spivak pessimistically comments “a full undoing of the canon-apocrypha opposition, like the undoing of any opposition, is impossible” (154). By looking at the shift in canonicity which is performed in Wikipedia and Wikidata we can be at least a little bit more optimistic: if there is no unifying canon, there is no static idea of what is canonical or apocryphal. In other words, Wikipedia reveals that a “Western canon” or global canon simply does not exist, at least for Wikipedia users.
National literature vs. world literature
The third aspect concerns the role of nationality in the world literature debates. It is one of the presumably irresolvable paradoxes that with the category of world literature, the category of national literature has experienced a peculiar revival. The various re-mappings of national literatures seem to be a delayed answer to Erich Auerbach’s apocalyptic vision:
Should mankind succeed in withstanding the shock of so mighty and rapid a process of concentration – for which the spiritual preparation has been poor – then man will have to accustom himself to existence in a standardized world, to a single literary culture, only a few literary languages, and perhaps even a single literary language. And herewith the notion of Weltliteratur would be at once realized and destroyed. (Auerbach 3)
Auerbach’s vision responds to the experience of having witnessed the effects of rabid nationalism in Germany, leading to Nazism. He registers a paradox that we know all too well in the post-national, globalized, planetary age: if nation as a category disappears, subsumed within the global, then national literature as a category risks disappearing, too. Were this to happen, however, the very concept of world literature would be emptied. In contrast, Wikipedia allows for a less apocalyptic view because it represents languages, not nations. The language editions, the small languages, their respective authors reveal a rich world of literature that the focus on national authors has so far obscured, and that are often elided from histories of national literatures. At the same time, and this is the opposite movement, the online encyclopedia strengthens and shapes national perception. Among other aspects, our special issue demonstrates that the representation of marginalized literary languages or groups does not depend necessarily on the size of the nation or language edition. It also examines how the image of a national literature is created through circulation in local and global markets and how this national branding is reflected in Wikipedia. Last but not least, it points to how Wikipedia articles highlight or obscure the relevance of the national or regional for a certain genre, as is the case with magic realism, described by Matylda Figlerowicz and Lucas Mertehikian in their article. Following Gisèle Sapiro, we can conclude that “the national is not systematically at odds with the international, the transnational, the supranational, or the cosmopolitan” (500) – in both the real world and the virtual world of the encyclopedia.
Challenges for research
Analyzing Wikipedia, neighboring projects like Wikidata, and other websites brings several challenges. Some of these challenges can be summarized under the rubric of accessibility. Some sources are difficult to access because they are neither available in standardized formats nor centrally listed. Even projects like OpenSyllabus do not provide data equally suited for all purposes. Anna Sofia Lippolis states that in her case study (Italian literature), she does not have access to syllabi that would help to draw a more complete picture, nor can she automatically process scholarly articles due to access limitations on publishers’ websites and difficulties in automatically linking them to other content (e.g., Wikipedia pages).
Understanding (and correctly leveraging) the complex relationships between the different Wikimedia projects (e.g., Wikidata or Wikipedia) is another challenge. For instance, finding out which information is available in (or derived from) which project, or how up-to-date information is can be quite arduous. Wikipedia and Wikidata category and class systems are particular obstacles. In both systems, the objects of interest are not necessarily assigned the same class (or category). In Wikidata, literary works could be linked via the “instance of” property to the class “literary work,” but often they are also classified as “book” or just “written work,” as noted by Paula Wojcik, Bastian Bunzeck, and Sina Zarrieß in their article: “the internal categorization in Wikidata is sometimes spurious. For example, the Holy Bible is classified as a literary work, while its parts (e.g., the New Testament) are not. These categorizations are at the choice of individual contributors, which can have wide-ranging repercussions for a computational analysis.”
Furthermore, both systems are hierarchical, having categories with many subcategories, and all of them are rarely relevant for the task at hand. This frequently requires manual work to include missing articles (or exclude irrelevant ones) and complicates the extraction of relevant data (sub)sets. Other systems, like the “shelves” in Goodreads, pose similar challenges. Another challenge for research is the changing nature of the Web, as described by Daniel Carrillo Jara: “a Wikipedia article could have one thousand words today, but three thousand the next week.” However, as he concludes, the analysis and the conclusions are valid because they “explain a concrete moment in the digital representation of Peruvian literature.”
Accounting for and processing changes (e.g., the Wikipedia revision history) is quite tedious, as shown by a 2018 dissertation that closely scrutinizes each and every individual edit between 2003 and 2015 in the German Wikipedia entry on the writer Walter Höllerer (Bronner). Yet a selective description of an article’s revision history can already be useful for tracing the evolution of an author’s reputation within Wikipedia, as demonstrated in a volume on Chinese-Sinophone literatures using the example of Taiwanese writer Li Ang. It is pinpointed that she “was lifted to the rank of ‘world literature writer’ on October 19, 2010” (Chiu 227).
Another complication is that pages in Wikipedia language editions may be translated wholly or in part from other language editions. It can be difficult at first glance to understand the provenance of each page, although many of the pages created through translation in recent years contain a tag to that effect. Finally, from the point of view of reception studies, it is challenging to understand how long visits to Wikipedia pages last, and whether they are read in toto, in part, or simply clicked on and then exited: while visits indicate access, they do not necessarily indicate reading.
Having successfully acquired a suitable dataset for analysis, its content can pose challenges as well. Some contributions focus on those challenges, particularly bias and representation. For example, little is known about how readers of literary works participate as writers of Wikipedia articles or contributors to Wikidata.
Melanie Conroy specifically focuses on the gender gap in terms of editors, numbers of articles, and length of articles dealing with female writers in Wikipedia and Wikidata. The problems with using the data for an analysis are the subject of the analysis itself. She also discusses other issues, for example, the “preference for the new and technological” which leads to an over-representation of “popular literature and science fiction”. Meanwhile, Daniel Carrillo Jara discusses reasons for the unequal representation of particular groups on Wikipedia, for example, their “capacity to engage in voluntary labor, economic conditions to spend time writing on Wikipedia, and digital literacy”. He cites previous findings that online encyclopedias reproduce biases and preferences, but he also notes that Wikipedia can provide different views from different communities on the same topic (in his paper, the Quechua Wikipedia and the Spanish Wikipedia on Peruvian literature), which can be both a challenge and an opportunity.
Other challenges include the large percentage of data that is ingested (or produced) by bots and typically not checked by humans, inconsistencies in spelling, for example, in birthplaces or dates, the understanding and analysis of articles in many different languages, and the proper assessment of the “relevance” of a Wikipedia page – an issue that has been discussed before (Hube et al.) but always needs to be reconsidered for the research question at hand.
The contributions to this special issue not only raise our awareness about these challenges but, most importantly, they also greatly expand our knowledge about them. At the same time, this issue is meant to stimulate the field, and suggest new paths forward. Wikipedia and Wikidata are constantly changing, and we are aware that these articles have ‘expiration dates’ at least in terms of the statistics they present. The data might change considerably, the system itself might change parameters, censorship might play a role in specific countries, altering reception and contributor data. In other words, the data relied on here is not intended to be valid sub specie aeternitatis: chances are it may be quite changed even in the space of a decade, or even by the time you read this preface. However, the research questions and critical analyses included in these articles show us a number of ways of investigating Wikipedia and Wikidata.
The five articles assembled in this issue offer a prelude to further research on what is probably the most ambitious encyclopedic project since Diderot and D’Alembert’s Encyclopédie. They are a status report, fixed in time, on the content of a rapidly changing digital environment. Far from exhausting the possibilities for analyzing the representation of world literature in Wikipedia and Wikidata, they are intended to provide an impetus, a stimulus, and a motivation for future projects. We hope this special issue highlights what a promising source Wikipedia and Wikidata are for literary studies, and we hope to see more applications, especially some that make it possible to monitor how things change in real time.
We would like to thank the authors for their contributions and close cooperation, especially in the last weeks before completion of this issue. Another big thanks goes to our five peer reviewers, Fabio Ciotti, César Domínguez, Sandra Folie, Evelin Heidel, and Mads Rosendahl Thomsen, who helped to make this issue what it is. We are very grateful to Andrew Piper for giving us the opportunity to publish this special issue of the Journal of Cultural Analytics and managing editor Katrin Rohrbacher for guiding us through the process. Special thanks go to Jonah Lubin and (again) to Evelin Heidel and Mads Rosendahl Thomsen for their critical comments and the continuous conversation during the completion of this issue.
Funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under Germany’s Excellence Strategy in the context of the Cluster of Excellence Temporal Communities: Doing Literature in a Global Perspective – EXC 2020 – Project ID 390608380.
Unfortunately, the publication of Hube et al.'s article did not go as planned. It was submitted to the newly founded journal Digital Literary Studies (https://journals.psu.edu/dls), whose first issue had just been published (vol. 1, no. 1, 2016). The article was successfully peer-reviewed and accepted for publication. However, a second issue of the journal never appeared, and the journal closed after the first issue. In order not to let any more time pass, the authors decided to publish the article as a pre-print on the arXiv repository, and to date this is the only citable version.