Why Map Literature? Geospatial Prototyping for Literary Studies and Digital Humanities

By focusing on the process of building A Map of Paradise Lost—a geospatial humanities text-to-map project that visualizes the locatable places in John Milton’s Paradise Lost— this paper addresses the question “why map literature?” and demonstrates how the process of research prototyping is in itself a form of knowledge production. Through a series of prototyping moments, we address how the different steps involved in building a geospatial humanities project can produce new knowledge about the fields it relates to: literary studies and digital humanities. The prototyping moments make arguments that advance our understanding of Milton’s Paradise Lost, approaches to data visualization for cartographic comparison in and beyond DH, and models for interdisciplinary collaboration. Resume En se concentrant sur le processus de construction d’une Carte du Paradis Perdu – un projet d’humanites geospatiales qui visualise, dans une carte a partir du texte, les places localisables dans le Paradis Perdu de John Milton – cet article aborde la question «pourquoi une litterature cartographique?» et demontre comment le processus de prototypage de recherche est en soi une forme de production de connaissances. A travers une serie de moments de prototypage, nous abordons comment les differentes etapes impliquees dans la construction d’un projet d’humanites geospatiales peuvent produire de nouvelles connaissances concernant les domaines relatifs a ce sujet: etudes litteraires et humanites digitales. Les moments de prototypage presentent des arguments qui ameliorent notre comprehension du Paradis Perdu de Milton, des approches de visualisation des donnees pour une comparaison cartographique «a l’interieur» et «au-dela» des humanites digitales ainsi que des modeles pour une collaboration interdisciplinaire. Mots-cles: cartographie litteraire; prototypage geospatial; communication savante; Paradis Perdu


Introduction
Geospatial humanities is a significant and rapidly growing branch of the digital humanities and constitutes the practice of applying Geographical Information Systems (GIS) and other quantitative technologies to the study of the representation of spatiality in texts, often to literary or historical content. A multitude of geospatial humanities projects involve geovisualizing literary texts. As a fairly recent research area, the contribution of literary mapping is still being established: why do we map literary texts? (Cooper, Donaldson, and Murrieta-Flores 2016, 9). Scholars have questioned the value of geospatial humanities projects for digital humanities and literary studies, often with inconclusive remarks (Piatti, 2016). This uncertainty is understandable in a nascent field that relies so strongly on digital tools and methods that are as diverse as they are rapidly developing; no wonder why it is difficult to single out their contribution to scholarship in an ever-changing medium. Allison Muri (2016) draws on Alan Liu's (2013) claim that "in the digital humanities, experimental studies are important, valid, and necessary trials as we test new methods in a still nascent field. We cannot proceed without experiments and testing of hypotheses. We also need to ask-and answer-what is the meaning of a literary GIS to literary studies and textual scholarship?" (Muri 2016, n.p.). Until we can define what geospatial humanities encompasses and set its boundaries in order to make overarching claims about its scholarly landscape, we ought to concentrate on a case-by-case exploration of the significance of individual projects to the research areas they relate to. This very type of examination itself can help define the field's parameters.
We address the question of the contribution of geospatial humanities to both fields that are involved in this particular interdisciplinary instance-digital humanities and literary studies-by looking at the prototyping process of building a literary map of John Milton's Paradise Lost. Although research prototyping may have some overlaps with other established forms of knowledge production, such as developing a critical edition of a work or writing an article or monograph, many of the processes, such as data collection and interpretation, collaboration, and platform design, demand a different way of approaching the text. At the core, the steps necessary for creating any of the above require engaging with the text directly, as well as with the cultural and historical materials surrounding the text. By adapting Alan Galey and Stan Ruecker's (2010) expression "how a prototype argues," to "how a geospatial prototype argues," we address how the different steps involved in building a geospatial humanities project from a literary text tackle the question of how geospatial humanities can drive digital and textual scholarship forward. This study is carried out by addressing specific prototyping moments-critical decisions about data gathering and structuring, as well as decisions about the features of the app-that demonstrate how the process of building is in itself a form of knowledge production that can grant new ways of engaging the material and imagining technical solutions to collaboratively visualize complex, multilayered, literary space.

A map of paradise lost
A Map of Paradise Lost is an open access online project that situates John Milton's Paradise Lost in its specific historical moment-the seventeenth century-published in a map-oriented culture that was at the peak of the development of a cartographic consciousness in Europe. In "Milton's Maps," Morgan Ng (2013) argues that there is a "tendency among current literary scholars, despite enormous interest in the ' cartographic imagination' in Renaissance writing, largely to ignore the texts' actual visual counterparts. To explain the textual form of Paradise Lost requires equally close attention to the images which permeated Milton's mimetic consciousness, even after the onset of his blindness." (2013,428). Ng (2013) points beyond the textual references that are typically read alongside Paradise Lost to actual maps, such as the map of biblical lands found in the paratext of Milton's own family map, a 1612/13 printing of the King James Bible-one with which Milton would have no doubt been familiar with over the course of his life and that, according to Ng (2013), influenced his spatial thinking about the places mentioned in biblical accounts and depicted in Paradise Lost.
Paradise Lost creates a rich and complex world that draws on multiple histories, with placial references from ancient antiquity, biblical accounts, and Milton's contemporary world, to name a few. Milton's allusions to place can be understood in critical notes of particular editions, where editors attempt to contextualize and explain the complex and multilayered references to places and their significance. Paradise Lost is the first project of its kind that grants visual access to the world of "geographical continuity" that permeated Milton's spatial consciousness. This worldview refers to the prevalent notion of historical sequence of a seventeenthcentury English audience, namely the conviction that biblical events, like historical ones, progressed on a linear spectrum of geographical continuity, meaning that the land that the Ottoman Empire occupied in the seventeenth century is the same land in which biblical and classical accounts took place (Ng 2013, 433). The historical maps included in the project are meant to evoke these worlds imagined by Milton.
The project is an application of Geographical Information System (GIS) techniques to the text of Paradise Lost that is meant as an exploratory tool for researchers, students, and readers to investigate the complex and multitemporal space of the epic poem. The map includes the unambiguous place-name references and explanatory excerpts from editorial notes that address the significance of the places mentioned. Every place name is contextualized in an accompanying passage (see Figure 1), with more than one passage for places that are mentioned more than once. Rather than just overlaying placial references on a modern map, the project attempts to visualize some of the many temporalities that are merged into the world of Paradise Lost, captured in the georectification (matching points on a map image with corresponding points on a map in a geo-  which place-name references alone might be enough to communicate a purpose, such as the movement of wind currents in a region, the humanistic representation of place in literary maps almost always requires contextualization. A point on a map that virtually means "this is Paris" for Paris may not tell us much. But to engage with place names in a humanities context often means to engage with place rather than space.
Tim Cresswell defines place as "how we make the world meaningful and the way we experience the world. Place, at a basic level, is space invested with meaning in the context of power. This process of investing space with meaning happens across the globe at all scales and has done throughout human history" (2015,19

Scholarly communication and literary mapping
The emergence of the digital medium has ushered an age of experimentation with the form of scholarly communication, from research to dissemination. In the digital humanities, this experimentation has taken many forms, including digital editions, online journals and encyclopaedias, dynamic databases, and software prototypes.
Digital editions generally provide content in more accessible and networked ways. authoritative reading and final statement on a subject, the moment of publication of a digital edition is quite fluid since it can be published iteratively rather than finally, in contrast to print editions. Namely, it "becomes a permanent but potentially always changing documentation of an ongoing examination and processing of the objects in question. In this way, the edition as a publication is a process rather than a product."  Let us return to our borrowed question at the outset: "Why map?" (Cooper et al. 2016, 9). As Sebastién Caquard (2011)  it also puts users in the position to use the tool in ways and for purposes that we have not envisioned yet.

Prototyping moment 1: Close reading meets map visualization
A generic distinction of sixteenth-and seventeenth-century works that describe places, such as travelogues or chorographies, is their insistent contextualization of place names at use. Scholars have traced many works from which Milton draws spatial references and the ideological connotations with which they are associated.  (183)  Lost, symbolize the postlapsarian world and the straying away from the "true" path. One thing that becomes apparent, even at a glance, is that most place names are, in fact, collocated with moralizing content. For example, Rabba, Basan, and Arnon (see The spatial grouping itself happens across all place names related to the broader category and references all place names associated with idol worship and pagan cults, like Nebo, Hesebon, Abarim, the Asphaltic Pool (grouped together as worshipers of Chemos) and Euphrates, Bethel, and Dan (collocated with the Hebrews' worship of the golden calf).
In cases where the place appears more than once in the text of the epic poem, each mention is treated separately and has a passage extract and moral collocation (Figure 3).

This prototyping moment is not meant to present a novel argument about Paradise
Lost; as demonstrated through scholarly and historical examples, the connection between platial references and moral valence already exists. Instead, by traversing these categories against a backdrop of different historical maps, this project is meant to grant imagine visualizing other literary frameworks or arguments arrived at through close reading, for example, or including more georeferenced maps to study them against.

Prototyping moment 2: Parallel map visualizations
Navigating multiple works at once has never been easier-a stark comparison is the labour of using a bookwheel, once a groundbreaking sixteenth century-invention in the form of a rotating bookcase to facilitate an easier way of reading more than one large book at once, to tools like Juxta McGann (2012) categories. Georectifying historical maps that, although imperfectly, but more closely capture these broad spatial categories is a more productive solution for this type of project. In an attempt to learn more about the geography that Milton references and the surrounding areas in these different contexts, and to more legibly navigate the multiple layers, we have built a functionality that allows to split the screen into up to four parts, where each part can be customized separately (Figure 4).

Prototyping moment 3: Building capacity for interdisciplinary collaboration in mapping projects
The process of creating the data for geospatial humanities projects, including defining the categories and variables, extracting and cleaning data, standardizing it across, and reshuffling categories for creating the most readable and useful  geovisualizations, is a meticulous process that requires revisiting the data many times, especially since we are carrying out all these steps manually for the sake of accuracy, since the content of Paradise Lost does not easily lend itself to automatic methodologies. In interdisciplinary collaborations, all parties bring their skillsets and the division of labour happens with respect to individual expertise. This means that updating, revising, and moving the data to fuel the map in a project like A Map of Paradise Lost, would necessarily require the humanist and developer (corresponding to the expertise of the two authors) to both be actively engaged in a stepwise process that may not always be a productive division of labour, and can actually impede the project from expanding in the future without the continuous active involvement of a developer in part of the process. Our final prototyping moment grants non-developers more autonomy in contributing to the project, while also planning for longer-term sustainability without having to rely on too many outside variables that increase the need for updates and iterative control. The solution to both aforementioned considerations is the data pipeline (see Figure 5).
Organized in a single spreadsheet, the data platform is a building block of the application. All required processing of the data is carried out using the standard build tools that are used in the development of the app; we do not use an extra database, an extra server where the database lives, or entry forms. The idea of the pipeline is that any collaborator can edit the content with a knowledge of how to use spreadsheets and the very basics of the git-based collaborative platform GitHub.
If a team member makes a change to a spreadsheet and makes a commit to push the changes to GitHub, this sets off the pipeline which builds the app, validates if the data is complete and in the right format, and then publishes it to the web server (Figure 6). Through validation, ideally, nothing can be broken in the process.
Humanists do not need to engage with the building process, pipelines, and other technical details that have already been developed, and can focus on the content. This division of labour raises the questions brought up scholars such as Stephen Ramsey (2016) about whether a digital humanist has to be a coder or not; for this paper and project, this inquiry is rather narrow since the iterative nature of the project invites digital humanists, but also early modern scholars, to share their expertise on Milton by suggesting other potential close readings and literary frameworks for interpreting the spatiality of Paradise Lost. By making it a more straightforward task to contribute to the project, we are working towards a less labour-intensive, steeplearning-curve model that is more productive and accessible. Rather than insist that all contributors must have advanced coding skills and a background in early modern literature in order to equally contribute to all aspects of the project, we build on one of digital humanities' strongest suits: that through interdisciplinary collaboration,

Conclusion
By addressing the prototyping process of A Map of Paradise Lost, we sought to offer an explanation of select prototyping moments in order to address the question "why map?"-essentially pointing to how the process of mapping is not unlike that of close reading, and that through data gathering and visualization, existing and novel interpretations of literature can be validated, expanded, and contested.
The digital medium also encourages a space for creativity and for experimentation with scholarly communication and methodology; for example, comparative work across texts and maps can be visualized side-by-side for more intuitive exploration and comparison across multiple timeframes and works. Finally, by building the data pipeline for interdisciplinary collaboration, we sought to provide a model for collaboration in which contributors with different technical skills can more readily contribute to a project, hoping to encourage a community of practice among literary experts that can essentially be adapted to other projects and their respective content. In DOIng so, we are also conscious of the iterative nature of digital projects and their ephemerality in terms of maintenance, thus thinking and prototyping towards sustainability.