Scholarship in digital textual studies often vacillates between two poles: the species of computational text analysis, sometimes called “distant reading,” occupy one pole, while text encoding and digital edition building, as might be characterized by the Text Encoding Initiative (TEI), occupy the other. For decades there have been calls for research that moves between these poles — for “scalable” or “zoomable reading” [Mueller 2012] [Cordell 2013] — as well as research that locates itself emphatically in the continuum, such as Alison Booth's “mid-range” reading [Booth 2017]. In a 2013 conference dialogue, Julia Flanders and Matthew Jockers dreamed of methods that would “mak[e] it possible to see both scale and detail simultaneously” because “micro and macro approaches are really two faces of the same thing.” [Jockers and Flanders 2013, 14, 18] We might even claim near-consensus among digital literary scholars that our research should bridge scales, and that the distant reading versus text encoding dichotomy is both intellectually and rhetorically impoverished. Nevertheless, in these repeated calls for a middle way we might also mark a lingering anxiety in the field that, however much we might wish for research that moves easily among analyses at different scales, in practice such movement is rare. Both practical obstacles of time and labor, as well as conceptual obstacles of commensurability, result in most projects unfolding largely at one scale or another.

In welcome contrast, Katherine Bode's A World of Fiction: Digital Collections and the Future of Literary History not only models how scholars might stage meaningful dialogue between scales of textual analysis, but also makes a forceful argument for the necessity of such work. The book ranges over a number of prominent computational methods, from quantitative data analysis to topic modeling to network analysis. Its heart, however, is the construction of “a scholarly edition of a literary system” which pairs a curated and annotated dataset of Australian newspaper fiction with a critical apparatus that “elaborates the complex relationships between the historical context explored, the disciplinary infrastructure employed in investigating that context, the decisions and selections implicated in creating and remediating the collection or collections, and the transformations wrought by the editor's extraction, construction, and analysis of that data” [Bode 2018, 53]. In this vein, A World of Fiction comprises both a monograph exploring her findings about Australian newspaper fiction and an extensive “curated dataset” available for download or exploration at http://cdhrdatasys.anu.edu.au/tobecontinued/.

Like printed critical editions, Bode argues, a scholarly edition of a literary system “presents an argument about the existence of literary works in the past based on the editor's interpretation of the multiple transactions by which documentary evidence of the past is transmitted to the present” [Bode 2018, 53]. Bode's edition, for instance, seeks to capture the shifting elements of fiction as it moved through the Australian press, such as changing titles and varying authorship attributions so that relationships among witnesses are visible and available for analysis [Bode 2018, 41–42]. In theorizing her computational work in book historical and bibliographic terms, Bode accomplishes something both transformative and refreshingly modest, demonstrating how computational or algorithmic models relate to the hypotheses underlying the work of all edition building, digital or not. The resulting book states its arguments and interventions plainly, while exposing its underlying evidence in an accessible interface, so that other scholars can verify, extend, or challenge Bode's findings. Though A World of Fiction declares its ambition “to advance a noncanonical, data-rich, and transnational history of the literary, publishing, and reading cultures of nineteenth-century Australia” [Bode 2018, 3] it is modest insofar as it refuses the rhetorical excesses that can attend other data-rich humanities projects, insisting instead on the createdness of its data and the situatedness of its findings.

Bode establishes the urgency of her approach in A World of Fiction's first chapter, which critiques the work of previous cornerstone books in computational text analysis: Franco Moretti's Distant Reading and Graphs, Maps, Trees and Jockers's Macroanalysis. Bode rejects the idea that the methods labeled “distant reading” or “macroanalysis” should be understood as opposites to close reading. Instead, she argues that both distant and close reading “are united by a common neglect of textual scholarship,” and “cannot benefit from the historical insights presented by editorial and bibliographic productions.” By “conceiving data and computation as providing direct access to the literary-historical record” such scholarship ends up “offer[ing] ahistorical arguments about the existence and interconnections of literary works in the past” [Bode 2018, 18–19].

Bode analyzes such oversights in detail, but in brief overview, she critiques the way such studies discuss the composition of their corpora (what they include and, most importantly, what they do not); the ways in which metadata can mislead (a single publication date, for instance, often obscures a more complex production history); or the ways such datasets elide broader publication contexts (the older books still on the market in a given moment, for instance, or parallel periodical publications). Bode takes particular issue with the assumption underlying many distant reading analyses that the date of work's first publication provides the most meaningful datapoint for modeling literary influence “in a chronologically discrete manner, regardless of the actual conduits of literary influence, which require availability to readers who buy, borrow, and sometimes write literary works” [Bode 2018, 29]. In arguing that distant and close reading are similarly ahistorical in orientation, Bode begins A World of Fiction with a striking and important claim. Taking that claim seriously would require practitioners of computational text analysis to pay more than lip service to the work of archive and collection construction, which Bode attempts to model in the following chapters.

In chapter 3, Bode establishes the centrality of newspapers to the publication and circulation of fiction in nineteenth-century colonial Australia; it was not one medium for fiction, but the predominant medium. Thus A World of Fiction asserts that its claims about newspaper fiction can be understood as claims about fiction in the region more broadly, more so than would be the case in a discussion of Britain or America in the same period. By relying on newspapers, Bode demonstrates her commitment to grounding A World of Fiction not only in the messy historical medium in which readers would have encountered these fictional works, but also in the histories of collection, preservation, and eventually digitization that make that newspaper fiction available for computational analysis, a process she outlines in detail in chapter 3, which details the social and technical structures of the Trove collections from which she draws her newspaper corpus. Importantly, Bode also notes the comprehensiveness of Trove's collections in contrast to similar digital newspaper archives in the US, such as Chronicling America, which lends even more force to the book's claims.

Part II of A World of Fiction transitions from laying theoretical groundwork to applying its theories to specific case studies, analyzing what the book's scholarly edition of a literary system indicates about authorship, anonymity, and reception of fiction in nineteenth-century Australian newspapers (chapter 4); what network analysis might show about circulation and the influence of syndication networks (chapter 5); and how topic models might reveal shifting regional and national literary traditions over the study's time period (chapter 6). At this point, I must confess one limitation in my own perspective. While I work extensively with digitized historical newspapers, my focus is on American (and more recently transatlantic) newspaper literature. Thus I feel most qualified to comment on Bode's methodological interventions in A World of Fiction. I am not expert in nineteenth-century Australian literary history, and so must take Bode's claims of intervention in this domain as read, assuming her characterization of the existing scholarship is made in good faith and that her findings challenge them in the ways she outlines. I will refer to these claims primarily as examples of how A World of Fiction moves between data and literary history, a move this book makes with particular alacrity. In reading, I learned a good deal about how nineteenth-century Australian newspaper fiction differed from American newspaper fiction, and in fact one additional merit of A World of Fiction is that it brings into focus a literary domain that is largely neglected in American literary study. Nevertheless, I must acknowledge that an expert in Australian literature might review some of Bode's claims differently.

In chapter 4, Bode works with the metadata she has assembled about the authors of Australian newspaper fiction to explore their demographics and reception in metropolitan and provincial newspapers. In some cases, A World of Fiction confirms previous scholarship, finding for example a trend away from anonymous authorship as the nineteenth-century progressed, though this trend is less definitive than prior narratives have claimed. From here, however, A World of Fiction advances a number of novel interpretations of authorship in the period. First, Bode finds that named authorship is far less important than attributed nationality. She writes, “47 percent of titles were inscribed with an authorial gender and 79 percent with a nationality. This relatively low rate of gender inscription…emphasizes the point already made: that the author was not the primary framework through which colonial readers interpreted fiction” [Bode 2018, 96]. Digging into gender further, Bode finds that, in contrast to much scholarship on nineteenth-century periodical literature, colonial Australian literature was more often written by men than women, and more likely to be local than is typically understood. Tracing trends across the century, Bode finds that prestige shifts over time: in urban newspapers, at least, local writing is valued more earlier in the century, when it is ascribed mostly to men, and loses prestige later in the century, when more women are contributing fiction to Australian newspapers. These are only a few of the trends Bode traces in this chapter, which demonstrates clearly how the data she collects and annotates can be used to elucidate the complex social and demographic situation of newspaper fiction in the period.

In A World of Fiction's next chapter, Bode explores the effects of syndication networks on the distribution and circulation of fiction in nineteenth-century Australian newspapers. She begins, however, with a strong caution of network analysis as it is typically implemented in data-rich literary history. In particular, Bode warns that “The probability measures needed to model systems based on highly incomplete datasets are at odds with the centrality of documentary evidence to historical argumentation.” While I am not entirely convinced by the definitive claim that “Literary historians focus on what occurred and why, not what might have taken place based on assumptions and probabilities” [Bode 2018, 130], Bode's overall skepticism about network graphs constructed from mostly incomplete and partial datasets is well taken, as is her critique of analysis focused on network visualizations rather than their underlying metrics. In the ensuing chapter, Bode uses network analysis primarily to explore her data rather than to make quantitative claims about it, and she does not employ network visualizations at all, instead charting the proportions of serialized and reprinted fiction across a number of categories: e.g. Australian vs. British, or metropolitan vs. provincial. Her findings demonstrate the importance of provincial networks to the publication and circulation of Australian fiction in the period, acting as a counterforce to dominant metropolitan syndicates that imported British and American fiction into the colony.

In its final content chapter, A World of Fiction uses a combination of topic modeling and decision trees in order to determine “whether, and if so what, characteristics distinguished the American, Australian, and British fiction in colonial newspapers.” As in her exploration of gender in chapter 4, here Bode finds that topics hew more closely to author's national identity than to their gender. The Australian literary voice is characterized most strongly by topics drawn “from prominent descriptions of nonmetropolitan colonial settings, characters, and activities,” but also by more “prominently depicting Aboriginal characters” than previous scholarship has acknowledged [Bode 2018, 159]. This latter claim requires exposition, and while Bode finds a much greater presence of Aboriginal characters than expected based on previous theorizations of colonial literature, she outlines how these depictions rely on negative stereotypes, such that the presence of Aboriginal people actually serves to justify the colonial project in most of these works. Bode notes that “In identity politics and literary criticism…oppression and emancipation are equated with nonrepresentation and representation, respectively” while she finds the opposite in the Australian newspaper fiction of the nineteenth century, which “asserted colonial ownership by prominently and consistently depicting [Aboriginal characters], sometimes in highly unsettling ways” [Bode 2018, 196]. Throughout this chapter, Bode moves deftly between the statistical relationships between words in her study's topic model, to the probabilities these topics would be associated with particular categories of author, to literary-historical interpretation that sometimes confirms, but often nuances or challenges existing scholarship.

In her conclusion, Bode overviews the interventions of A World of Fiction while opening up her discussion in a few pressing directions. First, she turns to the many questions the book did not answer, whether due to time or disciplinary expertise, and invites other scholars to engage with her scholarly edition of a literary system — in other words, with the curated and annotated datasets available as both downloads and in the interface linked to above — to engage those questions or formulate their own. Bode writes that the “scale of fiction encompassed by this scholarly edition and its international breadth make me confident that almost any historian of nineteenth-century literature will find something to extend their knowledge and enrich their research in this representation of a literary system” [Bode 2018, 205], a claim which seems sure to bear out as more scholars engage this work. Bode closes the book in a brief but urgent discussion of funding for resources such as Trove that are invaluable to the next generation of literary historical work but are in constant danger of defunding and loss of public support. Here Bode addresses the final but essential node in the literary system, the one which makes these kinds of analyses possible.

In all, A World of Fiction is an important work of scholarship that should be read, at the least, by scholars in digital humanities, nineteenth-century studies, book history, periodical studies, and library science. The book should certainly become a standard reference point in courses on humanities data analysis, cultural analytics, distant reading, and related methods. Bode brings the two poles of digital textual studies together compellingly, demonstrating how computational analyses might bring nuance and complexity to our readings of digitized archives of historical texts, and how the contours of those archives — the circumstances of their collection, selection, and digitization practices — might shape the framing of data-rich analyses. Her computational work is founded on and helps build a critical edition; the two activities are mutually imbricated rather than unfolding as distinct projects. Rather than restricting the work, Bode's firm disciplinary perspective ensures that the book's claims will resonate with area scholars: that the interpretive consequences of her computational work are clear. Bode's book productively challenges the dominant paradigms of “distant reading,” offering a more nuanced path forward for practitioners while inviting a much wider range of scholars to the data discussion.