Julian Schröter

A Mixed Methods Research Design for a Reader Response Based Analysis of Genre Expectations

Benjamin Gittel, Fiktion und Genre. Theorie und Geschichte referenzialisierender Lektürepraktiken 1870–1910. Berlin/Boston: De Gruyter 2021. 460 p. [Price: USD 163,86 / EUR 109,95]. ISBN: 978-3-11-073796-7.

One of the challenging aspects of Digital Humanities book publications is their hybridity concerning methods and their style of reasoning and writing. Benjamin Gittel’s book Fiktion und Genre. Theorie und Geschichte referenzialisierender Lektürepraktiken 1870 – 1910 (Fiction and Genre. Theory and History of Referencing Reading Practices from 1870 to 1910) exhibits its hybridity by making its mixed methods explicit. The monograph investigates how readers generate relations between fiction and the real world based on genre expectations. For this, Gittel includes four fields of research: (1) literary theory, (2) annotation-based, historical reader-response analysis, (3) descriptive and inferential statistical analysis, and (4) social literary history within the frameworks of praxeology and mixed methods research. While each field of research has developed its own style of reasoning, writing at the intersection of different fields is challenging because it requires authors to develop new styles. As reflected by the discussions on mixed methods research, this is not only a matter of presentation but also one of methodology. Therefore, this book review is intended to discuss not only the book’s impact on each of the four research fields, but to look at the ways the monograph integrates the different fields of research.

To prevent readers of the book from false expectations, I would like to clarify a potential ambiguity in the scope of Gittel’s book. Although it starts from a deliberate notion of praxeology and positions itself in the field of praxeological research, it turns out that its interest in reading practices focuses on cognitive procedures that have counted as heuristics or presumptions in hermeneutics. Recent scholarly perspectives, however, are often interested in literary practices covering even more practical actions including rigid activities such as prohibition, censorship, or public intimidation. While such a broader notion of practice is certainly relevant to late 19th-century culture, Gittel does not analyze actions beyond semantic expectations. However, with regard to the book’s particular interest in genre expectations, this narrow focus is well founded. Therefore, this note is not meant as a critique but as a hint to prevent readers from not having their expectations met.

The book is organized around its central hypothesis that there are genre-specific reading practices that establish the way readers relate texts to the world. The empirical verification and examination of this hypothesis are developed in ten chapters. The first four chapters focus on theory and methodology, while chapters five to seven present case studies on different subgenres of the late 19th-century novel (Historischer Roman, Tendenzroman, Zeitroman). A first synthesis relates the empirical results to selected historical perspectives (chapter 8). Chapters nine and ten can be considered as a two-part conclusion in the form of a pilot study on the Schlüsselroman (roman à clef) around the year 2000 and a summary of the results from the perspective of social literary history.

Following the composition of the book, I start with its theoretical part. The first chapter clarifies the analytical terms that will be used throughout this study. The second chapter establishes the praxeological and socio-historical framework, whereas the third chapter translates theoretical terms and relations to annotation categories. Lastly, the fourth chapter elaborates on a mixed methods research design. Two major goals seem to overlap in these first chapters: that of introducing theoretical macro models to be used in the subsequent analyses, and that of revising several of these macro models themselves within the field of literary theory.

Regarding the first goal and based on the concepts of social practice, social institution, rule, and convention, Gittel clarifies the concepts of literary practice and the literary institution. These clarifications are important for his analyses and valuable for socio-historical literary studies in general. Moreover, he provides a rational reconstruction of Fricke’s distinction between text type (Textsorte) and genre. Gittel’s definition of text types as generalizations over textual features in contrast to genres as generalizations over expectations can be regarded as a new and fruitful approach for research on literary genres. Remarkably, text type and genre are used here in an exclusively synchronic way. This contrasts with the more popular dichotomy between a transhistorical mode (Genette) or Schreibweise (Hempfer) that denotes types in the systematic description language versus genres as historical manifestations of text types. Gittel’s explication is consistent. It ignores, however, the need for historical models of historical change and the evolution of genres. As the historical perspective in the final synthesis (chapter 8) covers the change only with regard to the practice of fictionality but not of genres, the terminological choices may be sufficient for the book’s goals. However, the historical change in genre expectations is not covered by Gittel’s theoretical framework. It must be addressed and elaborated on in future work. Regarding the second theoretical goal of revising institutional accounts of literature and fiction, Gittel enters a complex discussion of literary theory without engaging in the details of preceding controversies. It seems to me that, while Lamarque’s and Olsen’s institutional account of literature is examined in detail in the subsequent case studies and revised from an empirical perspective, theories of fictionality are just used in the following analyses instead of being revised.

The third chapter derives a comprehensive schema of relations between fictional discourse and fictional worlds on the one hand, and real-world entities on the other hand. These relations include denotation, instantiation, and exemplification between propositions and facts, as well as between objects and classes. These abstract relations result in seven main types of references including object reference, instantiation, exemplification, class identity, explicit as well as implicit truths, and isomorphism. While not all possible sub-relations – which are listed in the book but not reported here – are eventually used in the subsequent empirical analyses, the main types are substantial for developing relevant annotation categories in the empirically central step of the book.

The fourth chapter embeds the process of annotating 363 historical reviews from the late 19th century on fifteen novels assigned to the four mentioned text types into a sequential explanatory mixed methods design. By sequential explanatory design, Gittel refers to a primarily hypothesis-driven quantitative approach including statistical analyses followed by qualitative interpretations and illustrative extensions that shall elucidate details and aspects that could not be answered by the quantitative study. The statistical analyses will not be evaluated in detail here. As the book analyzes only a limited range of data, discussions about best practices of statistical inference should be brought forward by professional statisticians. Gittel justifies his choice of different data and variable types, tests, and assumed distributions with exemplary diligence. Moreover, Gittel also considers possible – and more or less – reasonable and likely objections. In this respect, Gittel imports a virtue that is characteristic of analytic literary theory into the style of discussing empirical methodology and thereby provides a model of hybrid mixed methods writing to the Digital Humanities. I shall return to two relevant objections at the end of this review.

There is, however, one shortcoming concerning data collection and annotation that Gittel does not explicitly anticipate. Due to the limited research funds, it is understandable that annotations were not embedded in the historical reader-response documents and that these documents are not provided as machine-readable documents. Instead, the resulting annotations are made accessible only in a tabular format separate from the reader-response documents. Having machine-readable documents and the annotations on sentence level in a structured text format would be incredibly helpful for follow-up research within computational literary studies.

In the case studies of chapters five, six, seven, and nine, the initial hypothesis is differentiated into more specific sub-hypotheses which are assessed based on classical non-Bayesian statistics. These hypotheses, which are performed on each text type and genre claim that text-to-world-references are more or less likely to occur in historical reader response depending on genre expectations. Hence, Gittel examines the difference between reviews that assign a novel to a specific sub-genre and reviews that are neutral to sub-genre assignments. In the next step, he introduces a further distinction between text type and genre effects. Text type effects point at text-to-world references that are characteristic to all historical novels independently of the genre assignments of historical reviewers. Genre effects, in contrast, refer to the differences in the levels of text-to-world references by reviewers depending on the sub-genre assignments the reviewers are committed to.

From an overall perspective, the book succeeds in providing new and substantial insights as it detects specific footprints for each text type as well as genre effects within each text type. The footprints for the distribution of object reference, exemplification, truth, and isomorphism are cumulated in Fig. 47 (p. 308) and related to common hypotheses on fictionality and poetic realism in the synthesizing chapter eight. In the case studies on each text type, genre-specific effects are also verified. If, for instance, a work of fiction is perceived as a historical novel, Gittel shows that object references are more likely to be observed compared to a genre-neutral perception. Moreover, each genre has its most relevant reference type: For the Tendenzroman (roman à these) it is the implicit and explicit truths, whereas isomorphism is dominant in the genre expectations of the Zeitroman.

I shall return to the philological significance of these results at the end of this book review after discussing several methodological constraints that are in parts discussed by Gittel. Aside from some very hypothetical and artificial objections, two concerns are of more practical relevance. The main line of argument of the book implies an unambiguous causal relation between genre assignments and text-to-world references. According to a claim that has already been made by Walton in his seminal paper on “Categories of Art” (1970), perceiving a work of art within a category implies that readers expect specific features to be present in the work and base their aesthetic appreciation on either recognizing or not recognizing these features. Hence, the causal relation is expected to originate from genre expectation to the perception of features. However, both directions of causality are possible in principle. As Gittel points out, the inverse direction would be that of classifying a work as an instance of a genre based on previously observed features. Accordingly, Gittel models the assumed causal direction from expectation to perception within the frameworks of a causal theory of action and inference to the best explanation. However, Gittel considers the actual direction of causality to be eventually unverifiable based on quantitative analysis. Here, he brings the sequential mixed methods design into play. Looking at the reviews from a qualitative perspective is the only solution Gittel considers for verifying causality. The issue of causality is, however, and as far as I see, not consistently addressed in the subsequent qualitative sections.

Irrespective of the old belief that empirical analysis can only show correlation but not causality itself, I believe that there are options to model hypotheses on the causal relation also in quantitative research designs. I believe that this would have required a different strategy of data collection. If we knew reader responses based on genre assignments that were provided prior to the reader responses themselves (based, for example, on the genre assignments from paratext labels, which are excluded for good reason in Gittel’s design), it would be possible to look at the likelihood for both possible causal relations in more detail. As the book does not consider such a strategy, future work should add one further quantitative step to the sequential explanatory design in order to solve this issue.

The second objection that Gittel anticipates concerns type effects and the construction of the corpus in general. One may be surprised that Gittel claims to observe text-type effects although he only considers actual reader responses instead of the texts. However, there are two prior assumptions allowing him to raise his claim: Firstly, that the reviews are related to novels of one common text type, respectively. I shall not question this assumption here for the genres and text types considered in the case studies. Secondly, the case studies presume that readers’ responses are rational, by and large. Based on both assumptions, inferences from reader response to text type can in principle be assessed by subtracting genre effects. However, Gittel only considers genres that include the expectation of higher-than-average text-to-world-references. This makes it more than difficult to infer on average text-to-world-references for fiction independently of sub-genre assignments. As Gittel concedes, it would have been helpful to also take into consideration genres that could have less-than-average text-to-world references. For a methodologically rigorous and hypothesis-driven case study, it could be objected that text type effects are not operationalized in the best possible way in this study. This brings me back to the initial question and hence to the assessment of the overall style of reasoning of a book that includes different fields and research areas.

A particular challenge that Gittel’s monograph faces is to reconcile two contrary requirements, that of providing a comprehensive historical outline of genre practices on the one hand and that of justifying empirical findings with statistical scrutiny on the other hand. It is a remarkable achievement of Fiktion und Genre that the demand for a holistic historical account leads to only very few empirical inaccuracies and that most of these inaccuracies are made transparent. The book provides a courageous step towards a fine-grained and stepwise development without losing sight of the overall argument.

I shall finally address one possible objection that Gittel does not anticipate. Concerning the empirical results, I can imagine that some literary historians might find the above-mentioned footprints including the text type and genre-specific text-to-world distributions of references hardly surprising as far as these footprints confirm previous knowledge. This would, at least, be an objection that scholars working on quantitative methods regularly face. I want to emphasize that this objection would be inappropriate to Gittel’s findings. While the general text-type footprints indeed confirm previous knowledge, the confirmation based on new evidence is valuable in itself. Furthermore, the subtle differences in the level of genre expectations between genres, as well as between genres and text types, are highly interesting and need further interpretation. In this context, Gittel’s classical statistical approach could be complemented not only by more extensive hermeneutic and qualitative analysis but also quantitatively by algorithmic models accounting for the relative tendencies and changes. In other words, future research could use numbers not only to explore corpora and to perform statistical hypothesis testing but also to describe the subtle historical changes in genre semantics.

Acknowledgments

This contribution resulted from the project »A Mixed Methods Design for Computational Genre Stylistics and Unstructured Genres. Towards a Functional History of 19th Century German Novellas«‚ (project number 449668519), which was funded by the Deutsche Forschungsgemeinschaft (DFG) as a Walter-Benjamin Fellowship.

2022-12-14

JLTonline ISSN 1862-8990

Copyright © by the author. All rights reserved.

This work may be copied for non-profit educational use if proper credit is given to the author and JLTonline.

For other permission, please contact JLTonline.