Critical Modelling: From ‘Models of’ to ‘Modelling as’

We have always been modelling

Modelling is always already there in the scholarly practice of analysing the multifarious entanglements of literary phenomena. At times, the metaphors deployed in scholarly literature take up an idea of spatial movement, of pathways and entangled trajectories, that model dynamic processes of cross-functional networking and relationship building. In such a scenario, models help to shift our focus from factual descriptions to mediated expressions of resonance as well as the multiple temporalities and literary processes of signification, transmission, adaptation, and transformation. Furthermore, as Anita Traninger aptly remarks in her contribution ‘Modelling “Literature”’,

the notion of literature—and in particular that of literature in a global perspective—is an effect of modelling in that it is informed by processes of selection and subsumption that create the field that is supposedly denoted in the first place. It is an act of judgement (or a series of acts of judgement) that converts texts into literature, in particular in a historical or transcultural perspective.

At times, through a model, we deploy concept models to prompt literary-historical constellations, such as ‘writing’ and ‘reading’,¹‘In order to characterise the activity of reading, one can resort to several models. It can be considered as a form of the bricolage Lévi-Strauss analyzes as […] an arrangement made with “the materials at hand,” a production “that has no relationship to a project,” and which readjusts “the residues of previous construction and deconstruction” […] Another model: the subtle art whose theory was elaborated by medieval poets and romancers who insinuate innovation into the text itself, into the terms of a tradition […] The studies carried out in Bochum elaborating a Rezeptionsästhetik (an esthetics of reception) and a Handlungstheorie (a theory of action) also provide different models based on the relations between textual tactics and the “expectations” and successive hypotheses of the receiver who considers a drama or a novel as a premeditated action” Full reference in Zotero Library.‘canons’, ‘epochs’, ‘periods’, ‘genres’, ‘networks’, ‘communities’, infrastructures of ‘circulation’, ‘centre-periphery cleavage’, ‘national/world literature’, and/or literary ‘(poly-)systems’. From this perspective, it can be argued that literary history is ‘the predominant site of active, explicit, and mildly critical modelling’ (Traninger, ibidem). Research data management confronts us with the models we have deployed in literary studies as practical and analytical frameworks for data aggregation and reasoning about the manifold relationships, connections, and demarcations manifest in our documentary records. Despite this widespread practice, models are not yet widely accepted as critical concepts in literary scholarship Full reference in Zotero Library.

Conceptual models and theory

As a travelling concept Full reference in Zotero Library, theoretical reflection upon models and modelling practices have benefited from a wide range of intellectual investments in other disciplines and fields of research Full reference in Zotero Library. In literary studies, however, the usage of the term remains to a high extent unmarked, undertheorised, and colloquial Full reference in Zotero Library. In literary scholarship, models are commonly discussed in accordance with literary theory. The relationship between them is then mainly considered as illustrative or explanatory. Models function as tools to present, understand, and illustrate concepts from literary theory or a particular aspect of them (cf. Full reference in Zotero Library). Accordingly, Helmut Bonheim Full reference in Zotero Library stated in his Literary Systematics that ‘models are parasites on theory, and they are more modest. A model often maps out a section of what the theory is about – its special use is to suggest a limited set of analogies on which the mind can work’.

Functions and uses of conceptual models

In the typology put forward by Bonheim, four uses and advantages of models are discerned for the formation and communication of scientific theory: “cognitive” (through models, a certain theory or set of assumptions is made explicit and conscious), “heuristic” (in their capacity as search-and-find devices that assist in the co-production of (new) knowledge), “critical” (when models call attention to the wickedness of the assumptions behind them in a way that unmodelled theory, if such exists, rather conceals or leaves out), and “didactic” (models that show and tell, while stressing method-awareness of literary analysis) Full reference in Zotero Library. As is apparent from the above, models are associated in Bonheim’s ‘literary systematics’ mostly with top-down procedures, as much as their diagrammatic manifestations come to serve as ‘fallacy-finders and ignorance-spotters’ Full reference in Zotero Library.

Mediating models

Whether models constitute materialised arguments about (particular sequences of) literary theory or come to function as an application or illustration of it, they are considered as part of a pipeline,²See also the etymological sense of the classical Latin word ‘modulus’, OED Online, s.v. ‘module, n.’, 2021: www.oed.com/view/Entry/120688. which directs the interest in theory and method to the examination of literary phenomena. Models serve both as mediators and as triggers of (trans)formation in the process of theory demonstration, of hypothesis testing or, as practiced more explicitly in computational literary studies, of datafication. Formalised conceptual models act as intermediaries or ‘cargo carriers’ (cf. Full reference in Zotero Library) between theoretical problematisations, concepts and/or constellations, and the worlds of data they help constitute for a specific research purpose. More specifically, as Julia Gouvea and Sylvia Passmore aptly stated,

models are part of distributed cognitive systems that include cognitive agents, the phenomena they have identified, the questions they have, a model and the desired knowledge they intended for that model to help develop, as well as the larger scientific community within which those models are built and evaluated. Full reference in Zotero Library

Seen from this cognitive perspective, literary modelling is less expansive, prescriptive, or definitive than a theoretical framework. The term encompasses the dynamic process of iterative interaction between theory, model-making, and the communication of insights through the model’s final form. In their capacity to be iterative design tools, models make abstract concepts both explicit and practical, as they relate and relay critical thought on the ‘processes’, ‘relations’, ‘networks’, ‘communities’ etc. subjected to scrutiny according to the specific literary data they themselves help to constitute. The design iterations, data remediations, and explorations that co-produce the models are elliptical and non-linear. In model-making, a non-linear dialogue between the surfacing model and the ongoing conceptualisation takes place in an iterative and incremental method of continual refinement.

Models leak

When models structure recorded data in a theory-consistent way, they cannot help but ‘leak’, that is: attest to the constant dialectic that runs through literary modelling, make explicit the assumptions and decisions models are based on, and display the facets they (sometimes deliberately) disregard. As media for theorising and building ‘digital communities’, critical models are essential for understanding how our objects of research are embedded in (data and conceptual) models, even as they make or resist such formalisations; and how the manifold relationships, connections, and demarcations manifest in our documentary records might be transformed through modelling interventions. Examined critically, models surface from the status of substrate to reveal a continual tension between definition and uncertainty, selection and complexity, structure and messiness, functionality and sustainability, standardisation and difference.

The evolving articulations of a model about these critical issues are customarily discarded, once the model is presented in its ‘final’ or ‘representative’ form. The ability of models, however, to become opaque about the terms of their “coming-to-be” should remain of interest and concern for critical analysis. In parallel, the archival lacunae in respect to the critical modelling process, which involves drafting propositions, revising, and reconceptualising the model, might be managed by making provisions for a discursive equivalent. A critical protocol of a particular model might thus reveal and exploit its construction process, in as much as it actively confronts and attends to the disposition of a model as a vanishing mediator, rather than passively describing the model as a neutral or transparent medium. Given that literary modelling can have a ‘black box’ effect, such a protocol might co-facilitate the interpretation and critique of models by justifying the critical modelling process (e.g. attending to the principles of ‘grounded theory’ (cf. Full reference in Zotero Library) or ‘design thinking’ (cf. Full reference in Zotero Library), in order to make the models transparent, documented, and understandable.³The development of the ‘WE1S Interpretation Protocol’ arose from a similar concern. On this implementation, see the keynote lecture, ‘Humans in the Loop: Humanities Hermeneutics and Machine Learning’, presented by Full reference in Zotero Library, PI of What Every 1 Says at the DHd 2020 Conference ‘Digital Humanities between Modelling and Interpretation’ in Paderborn (March 2020): https://www.youtube.com/watch?v=lnfeOUBCi3s. Moreover, the protocol might destabilise and unseat the hegemony that it exercises as a model to something.

In the same vein, Katherine Bode has made a strong case for a ‘critical apparatus’ of curated datasets that serve as a basis for analysis, in order to reveal the hypothetical nature of the modelled literary system (or the particular part or aspect of it taken under scrutiny). In her pivotal study on data-rich literary history, A World of Fiction: Digital Collections and the Future of Literary History, Bode considers explicit engagement with modelling through a critical apparatus as a key mechanism ‘to recognise and represent the inevitably transactional nature of the documentary record’, as much as to assess and manage ‘the inevitable, sometimes radical, disjunctions that exist between knowledge infrastructure, the historical context it is intended to represent, and the data employed in that representation’. Indeed,

the critical apparatus demonstrates and manages those contested features by describing and justifying the editor’s engagement with the documentary record relating to a literary work, including with the inevitable gaps, remediations, and uncertainties that engagement exposes and creates. Full reference in Zotero Library

For (data-rich) literary history, the critical apparatus is thus integral to the ‘scholarly edition’ of a curated dataset and its underlying model, Bode concludes, because it

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. Full reference in Zotero Library

Modelling and/with literary data

With reference to scholarly work on particular objects of research, not least in the emerging field of computational literary studies, models became embedded more directly in research practice, whereas theoretical work—generally speaking—holds more to reflection. But as much as models are never self-sufficient, self-explanatory, or totally independent of theoretical preconceptions, they are equally grounded in the procedures of recording and transforming information into data (bottom-up).

“Models of”

Prominent accounts of the relationship between models and data maintain that scientific models are (albeit, idealised, distorted, simplified, abstract, and/or formalised) representations of “real-world” phenomena that the modeller is claiming to observe. In his classic study Allgemeine Modelltheorie Full reference in Zotero Library, Herbert Stachowiak, for instance, argued that the model is a representation of reality, despite its reduced complexity and specific focus. Such a strand of theoretical discussion about modelling insists on a representationalist epistemology, in which models represent and iteratively refine (our knowledge about) the empiric and material manifestations of literature. Sometimes, this representationalist stance even makes the model appear as a factual expression rather than acknowledging the modal and figurative foundations of the representation.⁴For an analysis of how representationalist accounts of modelling fail on both epistemological and ontological grounds, see Katherine Bode’s insight with reference to the work of Karen Barad Full reference in Zotero Library and Guilherme Sanches de Oliveira Full reference in Zotero Library. It is, therefore, important to keep in mind that the relation between both phenomenon and its formalised representation through a model is always selective, perspectival, and critically substantiated.⁵Katherine Bode quotes in this context from De Oliveira Full reference in Zotero Library to underline the specific material and normative conditions and contexts for the practice of modelling: ‘Models are “about” target phenomena as much as they are “about” the discipline in which they are used, the theoretical context they are meant to fit and advance, the methodological and technological background they are built upon, the intended users and the intended goals they are meant to help accomplish’. As such, models serve as a kind of depiction, imitation, template, paragon, or analogy: in short, a medium bound to form and representation.

“Models for”

In their introduction to the Springer Handbook of Model-Based Science, the editors Lorenzo Magnani and Tommaso Bertolotti put forward a functionalist definition of a model by stating that it is ‘something we [as scholars] use in order to gain some benefit in the understanding or explanation of something else, which can be called the target’. The notion of ‘benefit’ in this definition equally stresses that a model is targeted at something. Through modelling, this ‘something’ becomes bound to form, and the representation of our insights about it is, in turn, bound to a medium that assumes the function of a carrier of the representation. However, following Bernd Mahr, apart from being epistemic objects in themselves and the product of induction and/or deduction processes, models are purpose-bound too, not least in their capacity to be retargeted at other (similar) objects: ‘Ein Modell ist immer Ausdruck einer Möglichkeit, die sich einerseits im Modus der Aussage des Modellseins von etwas und andererseits im Modus der Aussage des Modellseins für etwas zeigt’ Full reference in Zotero Library.

Doing critical modelling

Thus, generally speaking, a model is something, about something, for something. However, the representational claim that models ‘show and hide’ seems rather to enforce than to diminish the reluctance to accept models in literary scholarship. On the one hand, models are often criticised for their acclaimed status as self-evident, impartial, self-sufficient, explicit, and sometimes even for having factual and/or universalist aspirations. Whereas, on the other hand, they are held to be unaccountable, reductionist, lacking transparency and legitimacy within a particular research design, and with limited flexibility to respond to a dynamic reality.

But let us not short-change the potential of models in literary scholarship by arguing against or focusing dogmatically (and rather predictably) on the reduction and abstraction, at the expense of their ability to revisit, reconfigure, or recharge research. In the light of the repeated criticism, the intermediary position of models between the ‘ideal’ world of theory and the complexity of reality or empirical phenomena seems to suggest that the conceptualisation of models and modelling in literary scholarship is more in want of a pragmatology than an ontology. Rather than conceiving models as devices for discovery and interpretation, understanding modelling in praxeological terms suggests paying closer attention to the affordances of models.⁶‘With this type of modelling as a translation of a contested concept into institutional structures, genres, and media, thus into a set of materialised practices, the question of discovery is replaced by that of affordances: which notion of ‘literature’ is afforded by these structures?’ Anita Traninger, ‘Modelling “Literature”’

Literary modelling is explicitly committed to ‘foreground[ing] the constructedness of knowledge and the observer’s place within it’ Full reference in Zotero Library and, in parallel, encompasses the ‘scholarly’ practice of ‘reasoning and thinking both about and with [literary] data’ Full reference in Zotero Library. The formalisation of meaningful attributions and the identification of properties to be recorded, together with the screening, enrichment, and interlinking of information in literary data, thus constitute the predominant interpretive and critical activity of modelling. Literary modelling is as much an argument about representation as it is about reading (as) data.

Obviously, there is a strong iterative and constructive dimension to critical modelling. Data and concept models implicate us as literary scholars within them; what we observe and analyse through (the construction of) models allows us to build sense – and this implicates us. By using the term “critical”, I thus refer to the self-reflective movement of the modelling agency, which is not absolute or transcendent in relation to the object of its modelling. The design of models is an integral part of our critical reflection on the constructed and transactional nature of the documentary record. We constitute the information that is remediated as structured data and we are reminded at every step of our research design that we are actually dealing with digital renderings of selected aspects in real-world literary phenomena, rather than the phenomena themselves. Models are a means by which we urge ourselves to become more explicit about the research questions we want to address and the data needed to process them. Models help clarify the cognitive steps that are involved in the pursuit of making literary-historical statements. Within the dialogical context of developing and testing a research question, models bring our own pre-conceived notions to the fore and establish a discursive site for positing this kind of research more explicitly in relation to the critical vocabulary and the infrastructure of knowledge-making.

The (epistemic) emergence of ‘temporal communities’

Modelling beyond representation?

The representational conception of models is perceptible in most critical thinking about modelling and data construction. While models do often designate and claim to represent entities and relationships, properties, and attributes of real-world phenomena, they could also perform other functions and inform other scholarly practices. These involve, for instance, (re-)calibrating the way we see and think about literary-historical phenomena (reading as); serving as a common frame of connection-building across objects and domains of research (reading by relay); and, last but not least, helping to conventionalise ways of perceiving and/or defining emergent features, which elude observation and cannot be recorded retrospectively or prospectively in a clear way, e.g. by mere description of entities and their relationships. As an outcome of such a critical modelling exercise, which is enacted beyond representation, ‘temporal communities’, for instance, emerge or may even become inflected (rather than being a timeless, stable, or self-evident object ontology which occurs in literary history as something waiting to be discovered) by the critical engagement and intervention of a literary scholar, interfacing and interacting at EXC 2020 Temporal Communities with the worlding effects of his/her literary models and data collections.

In their capacity to capture, orient, determine, intercept, frame, control, or secure literary phenomena in literary history, we may even come to consider models in literary scholarship in the discursive-strategic sense of a ‘dispositif’, as conceived of by Michel Foucault in his discussion of ‘the nature of an apparatus’; namely, as something

essentially strategic, which means that we are speaking about a certain manipulation of relations of forces, of a rational and concrete intervention in the relations of forces, either so as to develop them in a particular direction, or to block them, to stabilize them, and to utilize them […]. The apparatus is precisely this: a set of strategies of the relations of forces supporting, and supported by, certain types of knowledge.⁷‘Le jeu de Michel Foucault (entretien avec D. Colas, A. Grosrichard, G. Le Gaufey, J. Livi, G. Miller, J. Miller, J.-A. Miller, C. Millot, G. Wajeman)’ (1977), quoted from Giorgio Agamben Full reference in Zotero Library. See also Full reference in Zotero Library.

Seen from this perspective, our critical engagement with models may be described as the construction and investigation of apparatuses, together with incessant attempts to assess them critically.

This outlook on approaches to modelling beyond representation is intended to open up critical space not only for thinking about models differently, but mainly doing things with models differently in literary studies. When, in literary scholarship, models are acknowledged as having an argumentative scheme about which an argument for a given statement is construed, rather than acting as authoritative devices to record and describe phenomena, processes, and practices in literature, critical modelling will generate focused occasions of perception and reflection, will craft well-defined scholarly objects of research and contexts for interpretation, and will communicate a more explicit understanding about our procedures of coming-to-know. This is especially relevant because, as Katherine Bode Full reference in Zotero Library lucidly stated, when modelling, scholars actually ‘inscribe the boundaries’ of the object—that is, literally de-fine it as in delineation—that they ‘often presume to represent’.

“Modelling as”

Taking a cue from this theoretical insight, we might envision critical modelling and model-based reasoning as practices of literary scholarship, which focus not just on what is demonstrated, remediated, or contextualised through thick descriptions and analysis, but also in the background (i.e. the critical modelling) against which cognitive constructs, such as ‘temporal communities’, become utterable, knowable, and intelligible, in the first place. In looking at how models afford particular horizons of intelligibility, we come to realise how critical modelling is epistemology-oriented, rather than ontology-directed. It invites us to attend critically, not least, to the normative and naturalising authority exercised by models.

Maybe this insight becomes most tangible in a particular ‘mode of reading’ (to quote from Damrosch’s influential model for theorising ‘world literature’, Full reference in Zotero Library) which has gained considerable traction in literary scholarship during recent years: reading as. Recoined as ‘modelling as’, the formula gives the same opportunity for redefining the relationship between and the status of research object, researcher, and research question. In ‘modelling as’, we are again less concerned with the prevalent relationship between real-world phenomena and models-as-representation, but rather proceed by transposing and critically interrogating both resources—in short: the ‘model’ and the ‘world’—with equal status, i.e. as interdependent epistemic objects.

In this vein, ‘modelling as’ proceeds by interrelating and deploying an (immaterial) concept, as, for instance, a ‘temporal community’, through the prism of constructive comparison with the process of its materialisation and instantiation with data. What develops from the mutual conditioning of both the model’s assemblage process and research data production is the co-creation of critical models and ‘smart (research) data’ (cf. Full reference in Zotero Library), while acknowledging that both resonate with historical reality. Conceptual exploration and data construction are co-dependent on the performativity of the modelling process – that is, on the ways in which literary-historical concepts are embodied and enacted through (and thanks to) critical models. In doing this, they create, not least, a deeper understanding and demonstration about how concept-driven critical and literary-historical statements come to be articulated.

Critical models essentially afford an as-of-yet underexplored opportunity for coming-to-know, for construing and analysing. They serve as critical operators, both sites of construction and reading protocols, which ought not to be restricted to a modality of being or representing. The model functions for the literary scholar mainly as an intellectual grid that assists in the effort of locating the cognitive construct under scrutiny. It serves as a conceptualised transactional playground, which makes the coordinates and the viewpoint for her/his research more tangible. Rather than a prefiguration of real-existing data worlds, a normative reification of an assumed literary (poly-)system, or a descriptive object ontology of some historical reality conceived of as an ideational whole, the critically authored model is an active agent in the making of a space of intelligibility—the production and communication of knowledge.