In this important work, James E. Dobson addresses a long-standing criticism of the digital humanities (or DH): namely, that it is theory-free. The use of computers to work with large collections of texts is a hallmark of DH, and throughout this slim but dense volume, Dobson reminds readers that even the seemingly objective “data” extracted by computers from textual sources in fact encode (literally and figuratively) the assumptions and biases of their human makers and users. He argues, “What is needed for humanities scholars to use sophisticated computational tools is not a disruptive break with the past in favor of any particular ‘distinctive’ approach but instead a critical digital humanities—a vocabulary and framework constructed around a set of contested concepts and methodologies rather than a tacit agreement in terms of approach and objects” (5).
The first chapter lays out Dobson’s central claim that no scholarly methodology, computational or otherwise, is truly objective. The second considers some specific approaches within DH, including topic modelling, “a way to organize a large and unlabeled collection of documents into computer-generated ‘thematic’ categories” (49). In this chapter Dobson’s critique becomes particularly trenchant. He notes a scientistic turn in recent humanistic study, which he sees as “part of a retrograde movement that nostalgically seeks to return literary criticism to the structuralist era, to a moment characterized by belief in systems, structure, and the transparency of language” (57).
Chapter 3 argues for a careful attending to the historicity of texts. “Because of the large and rapid increase in size of these queries and objects, digital projects have the worrisome ability to draw attention away from certain questions by fetishizing the archive and imagining its objects as mere repositories of the past” (67). The chapter ranges over historicist and new historicist approaches and how these might apply to DH, and includes a lengthy discussion of sentiment analysis, a method for quantifying opinions expressed in texts by assigning positive or negative values using pre-made “dictionaries.” Dobson argues, “Sentiment, affect, feeling—whatever it is called and however it is theorized, has value to the humanities and especially to the study of literature, but depending on scored, tabulated, and averaged values of reported feeling in interpretive acts, machinic or hermeneutical, ignores the contingency of history and the situatedness of human culture” (99).
The fourth chapter considers the history of a particular algorithm, called k-Nearest Neighbor, or k-NN. Dobson’s premise here is simple and timely: “Understanding the scope, utility, and function of the rapidly multiplying algorithms used by businesses and governments that are hidden, embedded, or present in almost all areas of everyday life is necessary for any account of how culture and power work in the present moment” (102). In charting the history of the k-NN algorithm, Dobson shows how it literally encodes the concerns and biases of the historical moment in which it was created.
The book’s brief conclusion considers the potential applicability of neural networks to humanistic research. Dobson points out some limitations of newer approaches and points toward some possible methods to account for them, although they are not fully developed here.
Recent interest in “computational folkloristics,” situated explicitly in the DH universe, stands to benefit from the types of critical engagement and methodological reflexivity Dobson advocates. The book is not a handbook of method, and assumes a level of familiarity with coding, text-mining, and other computational approaches that limit its usefulness to newcomers to DH. But its powerful critique of quantitative methods in humanistic research is worth considering by folklorists whose work centers on texts and archives.
The book raises some important questions for scholars who, like this reviewer, work in broadly digital areas, but not with the quantitative/computational methods Dobson discusses. For example, when Dobson argues that “while computationalism and the use of digital methods in humanities work can render the study of culture scientific and structural, only a critical digital humanities can foreground cultural and epistemological questions about computationalism” (6), is he equating all of “culture” with written texts? How might ethnographic approaches articulate with the quantitative ones he outlines, beyond the simple translation of grounded theory (not typically in the wheelhouse of folklorists anyway) and into automated/computerized processes? How, in concrete terms, can quantitative approaches avoid the traps of ahistoricism and the “fetishizing” of the archives that Dobson rightly warns against?
These questions are not indicative of flaws in the work, however, but arise instead from its deliberately limited scope as an initial intervention. Critical Digital Humanities is an important corrective to approaches that frame computational data as inviolable, unbiased, and essentially trustworthy. Dobson masterfully combines cultural theory with complex computational approaches, and points to areas for further development by digital humanists looking to resist the slide toward scientism, remembering that they are, after all, humanists.
--------
[Review length: 791 words • Review posted on August 27, 2020]