“Open Source Interpretation Using Z-axis Maps” by Alex Christie and the INKE & MVP Research Group

1. Z-axis research: literary maps for theoretical interpretation

While geospatial methods in the humanities currently enable the visualization of place (mapping where events in a given text occur) such methods might still undertake the visualization of space (mapping how a text describes multiple places in relation to each other). Z-axis research is a 3D geospatial project conducted through Implementing New Knowledge Environments (INKE) and the Modernist Versions Project (MVP), with work ongoing by I and Katie Tanigawa in the Electronic Textual Cultures Lab (ETCL) (funded by SSHRC and in collaboration with Compute Canada). The project takes as its two main premises that (first) GIS-specific maps rely on contemporary understandings of geography often at odds with those expressed in literary and historical documents and that (second) cultural heritage maps express those historical understandings of space, to which literature of the same period responds. Proceeding from these two premises, z-axis maps anchor geographic data taken from modernist novels in contemporary cartographical expressions of the cities in which they are set, affording historically and culturally specific forms of geospatial expression and interpretation.

Z-axis mapping draws inspiration from alternative mapping methods developed during the modernist period, particularly the Situationist practice of Psychogeography, which cut-up and combined fragments of different maps to design imaginary and alternative pathways through urban space. Such pathways are often depicted in literature of the period, which (as the maps reveal) often warp and transform actual geography to represent subjective or interpretive accounts of them. Extending modernist artistic and literary practice into electronic environments, the z-axis project joins the interpretive turn in digital geospatial expression, including mapping platforms such as Scholars’ Lab Neatline and Hypercities, while also complimenting and revealing a range of activity across aligned disciplines, notably within the field of cultural geography. At the same time as the z-axis project thinks across divisions of literary and geospatial representation, it equally uses such thought to construct perspectives on divisions of gender, class, and sexuality as represented in modernist literature, resonating with Tim Cresswell’s statement that “The geographical ordering of society is founded on a multitude of acts of boundary making—of territorialization—whose ambiguity is to simultaneously open up the possibilities for transgression.” (149) A detailed analysis of the project’s theoretical engagement with modernist theories of space can be found in “Mapping Modernism’s Z-axis: A Model for Spatial Analysis in Modernist Studies.” (Christie and Tanigawa 2016)

Figure 1. Z-axis map of Virginia Woolf’s Mrs. Dalloway

In order to produce contextual maps, the z-axis project digitizes cultural heritage maps of modern cities and then warps and deforms those maps in 3D according to the z-axis data provided for each georeferenced place in a literary text. The “z-axis” variable measures the amount of description per text-unit (paragraph) in the text as an additional, contextual variable, in addition to x-axis (latitude) and z-axis (longitude) data. This creates a relief map of the distribution of a literary text according to the geography in which it is set (visualizing how much text the novel devotes to a given place). This relief map is then applied to a historical map that shows historical information, such as population density, wealth distribution, and zoning. Applying the third, z-axis variable to historical maps therefore adds a set of data that adds spatial context to place-based x and y coordinates for a given place, adding both visual and interpretive depth to maps that are otherwise flat: “Broadly defined, then, the z-axis refers to the cultural contexts for literary and historical data, advocating the inclusion of such contexts in data visualization.” (Arbuckle and Christie 2015) Although initially undertaken to explore literary representations of Paris during the interwar period through handmade maps, z-axis mapping was subsequently conceived as a method available to wide communities of scholars through the production of the z-axis mapping tool. This paper unpacks the automated production of z-axis maps and outlines how the findings of a given constellation of maps speak to each other, defining this process through the concept of “open source interpretation.” After laying out the z-axis tool and the findings it affords, this paper explores the concept of open source interpretation as an ecosystem in which multiple instances of new media scholarship (in this case, z-axis maps) speak to each others’ findings.

2. The z-axis mapping tool

In order to let scholars create, share, and comment upon z-axis maps, the INKE-MVP team is developing a z-axis mapping tool in collaboration with Compute Canada at zaxis.uvic.ca. Written in Python and Javascript, and hosted on the Compute Canada Westgrid servers, the tool allows scholars with minimal 3D modeling and geospatial expertise to use z-axis mapping to interpret texts in their own areas of domain expertise. It is being developed by Colin Jones, Katie Tanigawa, myself, Belaid Moa, Daniel Brendle-Moczuk, and Stephen Ross, with additional input from members of the INKE and MVP research groups. Currently, the tool allows the automated production of warped, 3D maps of novels set in London. The mapping process is divided into four phases: text input, NER correction, map visualization, and advanced settings. In the first phase, users are able to input texts into the tool by uploading a plain text file, pasting raw text into a text box, or choosing from a selection of pre-supplied modernist novels (options include E.M. Forster’s Howard’s End, Joseph Conrad’s The Secret Agent, and Virginia Woolf’s Mrs. Dalloway, among others). The text is then parsed in order to produce a list of place-names for the novel. The tool parses the text using Stanford’s Named-Entity Recognizer (NER) in order to produce a list of place-names in the text file; these pace-names are then georeferenced using a gazetteer of place-name coordinates for London. This parsing and georeferencing process produces a list of each explicit place-name mention in the text, associated with geospatial coordinates that identify the place (as well as a record of where the mention occurs in the text, with paragraph as unit of measure).

Once the user has selected a text and that text has been parsed, she is then presented with a screen for modifying the list of georeferenced place-names. This screen includes a list of each recognized place-name, the number of times each individual place-name is mentioned in the text, and an option to reject identified locations that the user does not want included in the map (users may also undo their rejections). After the user has filtered the tagged locations produced by the parsing and georeferencing process, she may then continue to the warped, 3D map of those filtered locations.

Figure 2. Parsed geo-data for Mrs. Dalloway

In the third phase, the user may view the z-axis map of the text and filtered geodata produced in the previous two phases. Users may choose between four georectified base maps from different decades: 1889, 1922, 1930, and 1966. These maps include contextual information as part of their cartographic design; for instance, Booth’s London Poverty map displays the relative income for each city street according to color codes. The z-axis data for the text being mapped is then applied to the historical base map of the user’s choosing, warping the map according to each georeferenced location. While the radius for each instance of warping is the same, the degree (or height) of the warping effect is determined by the number of times each location is mentioned in the text. The visual effect is produced by creating a conic shape underneath the base map in OpenGL, rather than manipulating individual pixels in 3D (a computational process far too expensive for present web applications). This warping method allows the tool to produce 3D maps, running in OpenGL, that consume relatively few computational resources and can be produced and visualized on the client side, in-browser. These maps visually anchor geographic data for a given text in historical maps that offer contexts within which that data can be interpreted.

Figure 3. Z-axis map of Mrs. Dalloway

The image above (fig. 3) expresses z-axis data for Virginia Woolf’s Mrs. Dalloway through Booth’s Poverty Map, showing the novel tends to privilege middle-class, or “well-to-do” areas, depicted in red on the poverty map. The z-axis tool offers further historical maps that users can compare against the z-axis map, including maps of London death rates, Zeppelin & Airplane bombing locations from 1919, and Population Density maps for 1901, 1911, 1921, 1931, 1939, and 1951. These comparison maps allow users to explore complex cultural connections between the same geographic coordinates, considering, for instance, if there are meaningful connections between the well-to-do areas described in Woolf’s 1922 Mrs. Dalloway and locations targeted for bombing during World War I. Users can also dynamically visualize how geographic references change as the reader progresses through the novel, animating the map in time by selecting a word range for the visualization. Each location parsed by the tool is associated with the paragraph in which it appears in the novel; the distribution of these paragraphs across the entire novel spans the text range from 0-100%. In this way, when a user selects 25%-50%, she is able to see which locations are mentioned across the paragraphs that span the second quarter of all paragraphs in the novel.

Figure 4. Z-axis map of Mrs. Dalloway, filtered by word range

This feature lets users dynamically visualize how the space of London shifts and changes in time throughout the novel, using the tool to explore connections between time and space. For instance, a user might visualize whether or not locations move to increasingly wealthy or impoverished neighborhoods as a novel progresses.

Figure 5. Advanced Settings for the z-axis tool

The tool also includes a range of advanced settings that lets users modify the z-axis map. In this, fourth phase, users can upload their own map and coordinates file to use as the warped base map in the tool. They can also adjust the peak height scaling, selecting among linear, logarithmic, and quadratic scales. (The default scale for the maps is logarithmic, which compresses the wide range of z-axis data such that very large and very small z-axis variables do not appear vastly different in size.) Additionally, users can set the minimum reference counts to create a peak, choosing to visualize only locations that are mentioned a certain number of times. Combined, these advanced settings allow the user to intervene in the visual display of the z-axis data for a given map, considering data expression as part of the interpretive and contextual map she produces.

3. Z-axis interpretation and the question of scale

The z-axis mapping tool is a continuation of existing z-axis research conducted by Katie Tanigawa and I with the INKE and MVP research groups.[1] This first phase of the project produced warped maps of Jean Rhys’s Quartet and Djuna Barnes’s Nightwood created by Tanigawa and I, respectively. Made by hand, these maps were produced by manually georeferencing each novel in xml-tei and warping a historical map of Paris in Autodesk Mudbox. The workflow and findings of this research are laid out in “Z-Axis Scholarship” (Christie et. al. 2014), “Arguing Through Archival Objects” (Christie 2015), and “Intersections Between Social Knowledge Creation and Critical Making” (Arbuckle 2015). These publications additionally discuss the theoretical issues at stake in this research, including z-axis method’s relevance to literary modernism. The z-axis maps of Nightwood and Quartet, while producing interpretations of each novel individually, further enabled comparative readings of modernist Paris that crossed both maps and novels (Christie and Tanigawa 2016). Opening this process of comparative cartography to communities of literary scholars is the chief aim of the mapping tool. To discuss the comparative findings the tool is designed to enable, this paper will now outline the findings of the project’s initial phase.

In Nightwood and Quartet, Djuna Barnes and Jean Rhys respectively describe the experience of living on the social and economic margins of 1930s Paris. To anchor the z-axis data for both novels in such economic contexts, the z-axis maps of these two novels use an interwar tourist map that privileges tourist locations over the common (or populaire) areas often frequented by Barnes’s and Rhys’s characters. This means that instances of warping re-expand impoverished locations that were shrunk for the original tourist map, visually expressing the tension between wealthy and impoverished visions of Paris in 3D. (Christie et. al. 2014) Barnes’s Nightwood focuses on the experience of lesbian and homosexual characters who live in Paris’s Latin Quarter. The z-axis map of Barnes’s novel visualizes the role class plays in the queer experience of 1930s Paris.

Figure 6. Z-axis map of Djuna Barnes’s Nightwood

Each instance of warping on the z-axis map corresponds to a geographic area described in the novel, with the radius of the effect corresponding to the specificity of the geographic reference. The map (fig. 6) reveals a Paris highly divided by class—references tend to be more specific on the wealthy right bank (north of the river), while the left bank areas inhabited by Barnes’s queer characters (south of the river) are often referenced vaguely. These areas are not referenced vaguely because the reader does not know where the actions takes place, but rather because the reader is not supposed to know where events are unfolding. Queer characters of Barnes’s day used the anonymity provided by public spaces to anonymously express queer desire without being identified, located, or targeted. (Christie and Tanigawa 2016). From this perspective, the z-axis map visualizes the wealthy right bank as a zone of sexual confinement and constraint and the left bank as containing zones of possibility, indeterminacy, and freedom.

Creating additional maps set in the same city and decade begins a collective investigation of the trends and patterns in a given city as described by literature of the period. While Djuna Barnes describes a poor Paris that offers spaces of sexual freedom, Jean Rhys describes a Paris that traps and torments its impoverished inhabitants. Tanigawa argues that, unlike Barnes’s characters who roam the Latin Quarter indeterminately, Rhys’s protagonist, Marya, hops among multiple specific bars and cafes in her attempt to find a place of belonging on the margins of Parisian society. (Christie et. al. 2014).

Figure 7. Z-axis map of Jean Rhys’s Quartet

While the Nightwood map visualizes zones of sexual freedom and liberty, the Quartet map expresses sexual exploitation and imprisonment in the interwar period. Taken together, these maps express different social and cultural currents in Paris during the interwar period; they are not mutually exclusive but instead call for a process of mutual interpretation. This collaborative interpretation functions through our written analyses of the z-axis maps, as well as through the maps themselves—the warped aesthetic of the Barnes map and the fragmented and disjointed aesthetic of the Rhys map both enable the spatial interpretations Tanigawa and I put forth.[2]

When read alongside each other, the z-axis maps of Nightwood and Quartet begin to address a research question that can only be answered at scale: how does modernist literature characterize the geographic experience of poverty in interwar Paris? This question requires the work of a community, including digital humanists with geospatial and data visualization techniques, as well as literary scholars with deep content expertise on authors and novels contemporary to the two mapped here. There is no singular graph or visualization that accesses the truth of impoverished Paris in the interwar period, since that experience can only be addressed through trend and pattern; just as a graph charts trends across multiple points, so too can constellations of multiple z-axis maps chart patterns in modernist geospatial expression.

Scholars see interpretations operating at scale every time they attend a conference or read a special issue or edited collection, and interpretive visualizations can operate at scale by taking the same social knowledge approach already present in traditional humanities venues. (To be sure, this is but one possible way in which visualizations might scale; there are many forms this process may take and all merit exploration and investigation.) This open source model for interpretation invites scholars to scale a project by addressing theoretical issues that can only be answered at the level of the disciplinary community. It conceives scale or pattern as a question of theory and interpretation, leaving further forms of analysis open for scholars to engage through a process of social knowledge creation.[3]

4. Open Source Interpretation: z-axis research as a social knowledge enterprise

The z-axis tool is designed to enable the large-scale production of z-axis maps, allowing communities of modernist and geospatial scholars to address questions of scale and trend in modernist literary expression. In order to facilitate this process of social and open interpretation, the team is expressly developing the tool to produce multimodal content that scholars can integrate into online and web-based publications. This will allow scholars to display warped 3D maps as part of online publications on z-axis research. It will further allow scholars to respond to each others’ z-axis interpretations through the tool itself, editing a scholar’s map and workflow to produce a different version of that map that responds to its interpretive display. In this way, z-axis maps are designed to function in a digital ecosystem of new media scholarship where scholars can directly experiment with and extend upon digital geospatial research. This process applies an open source model to scholarly interpretation—it makes z-axis materials, techniques, and findings openly available so that scholars can use shared research sources to either reproduce or respond to each other’s interpretive findings.[4] Doing so scales up interpretive discovery by situating it as a social knowledge enterprise. Here, the question of how humanities data functions at scale is simply a digital revisitation of the traditional task of uniting multiple scholarly voices in a given disciplinary community.

Through this social production of z-axis maps, the INKE-MVP team is using the open source z-axis tool to turn z-axis research over to scholars and scholarly organizations, such as the modernist studies community. Recasting z-axis mapping as a collaborative and community-driven endeavor allows the team to address a key research question in modernist literary studies that can only be answered at scale: how does modernist literature characterize the geographic experience of London in and around World War I? This, in turn, leads to still larger comparative questions, including what meaningful differences exist between the literary representation of urban poverty in Paris and London in the 1930s.The answers we produce collaboratively—expressed through a series of maps, datasets, and interpretations—will be open, in turn, for other scholars to change, modify, and expand. The scale of our guiding research question therefore goes hand-in-hand with a corollary scale of knowledge production and dissemination, through conference workshops, online publications of our findings, and openly disseminated, interpretive data visualizations.

The shareable and scalable z-axis findings produced using the tool express what Alyssa Arbuckle and I refer to as “self-sustaining knowledge production chains.” (Arbuckle and Christie 2015) These chains form when the open source research output of one scholar creates material for other scholars to reuse and repurpose in the creation of new data and findings (which both respond to the original theory while leaving the door open to still new research production). Such knowledge production chains traverse disciplinary and institutional boundaries, as the z-axis map produced by a modernist literary scholar may be of interest to cartographers, geographers, desktop fabricators, and so on; similarly, the transformation of cultural heritage materials offers multimodal output for hybrid publications, interactive visualizations, and the display of library collections. These cross-disciplinary production chains, and the scalable research questions through which they operate, are the fruits of open source interpretation. By repositioning research replicability at the site of interpretive methodology, open source interpretation affords a model of scale in which interpretive maps may be socially modified, remixed, and transformed. Ultimately, data visualization can scale by socially expanding the interpretive methods that shape and grow humanities datasets. Just as z-axis maps reveal social currents in the modernist city, so too does social knowledge creation cultivate and expand humanities research through trends and patterns in scholarly findings.

5. Notes

[1] Z-axis research began as a student project by Katie Tanigawa and myself in Jentery Sayers’s digital humanities course. Following the completion of our student prototype, the project continued in the Electronic Textual Cultures Lab and the Maker Lab in the Humanities, across the INKE and MVP research groups. Colin Jones, Belaid Moa, Stephen Ross, Jentery Sayers, Katie Tanigawa, and I comprise the full z-axis team.

[2] For further reading on the z-axis mapping project, see Christie, Alex and the INKE-MVP research team, “Arguing through archival objects: a z-axis method for 3D printed interpretation” (2015) at http://web.uvic.ca/~mvp1922/arguing-through-archival-objects-a-z-axis-method-for-3-d-printed-interpretation/; Christie, Alex, Stephen Ross, Jentery Sayers, Katie Tanigawa, and the INKE-MVP research team, “Z-axis scholarship: Modeling how modernists wrote the city” (2014) at http://web.uvic.ca/~mvp1922/z-axis/ and http://maker.uvic.ca/dh14/.

[3] For further information on social knowledge creation, see Alyssa Arbuckle, Nina Belojevic, Matthew Hiebert, Ray Siemens, with Shaun Wong, Derek Siemens, Alex Christie, Jon Saklofske, Jentery Sayers, and with the ETCL and INKE Research Groups, “Social knowledge creation: Three annotated bibliographies” (2014).

[4] For more on open source development, and its applications beyond software environments, see “What is open source?” at http://opensource.com/resources/what-open-source.

6. Works Cited

  • Arbuckle, Alyssa, Nina Belojevic, Matthew Hiebert, and Ray Siemens, with Shaun Wong, Derek Siemens, Alex Christie, Jon Saklofske, Jentery Sayers, & the INKE & ETCL Research Groups. 2014. “Social knowledge creation: Three annotated bibliographies.” Scholarly and Research Communication 5.2. http://src-online.ca/index.php/src/issue/view/16.
  • Arbuckle, Alyssa and Alex Christie, with the ETCL, INKE, and MVP Research Groups. 2015. “Intersections Between Social Knowledge Creation And Critical Making.” Scholarly and Research Communication 6.3. http://src-online.ca/index.php/src/article/viewFile/200/426.
  • Barnes, Djuna. Nightwood. London: Faber & Faber, 1936.
  • Christie, Alex. “Arguing Through Archival Objects: A Z-Axis Method for 3D-Printed Interpretation.” Last modified 16 January 2015. http://web.uvic.ca/~mvp1922/arguing-through-archival-objects-a-z-axis-method-for-3-d-printed-interpretation/.
  • Christie, Alex, Stephen Ross, Jentery Sayers, Katie Tanigawa, and the INKE-MVP research team. “Z-Axis Scholarship: Modeling How Modernists Wrote The City.” Last modified 1 August 2014. http://maker.uvic.ca/dh14/.
  • Christie, Alex, Colin Jones, Belaid Moa, Stephen Ross, and Katie Tanigawa. “Z-Axis Mapping Tool.” Last modified 15 Oct 2015. http://zaxis.uvic.ca/.
  • Christie, Alex and Katie Tanigawa. “Mapping Modernism’s z-axis: A Model for Spatial Analysis in Modernist Studies.” In Reading Modernism with Machines: Digital Humanities and Modernist Literature, edited by Shawna Ross and James O’Sullivan. Palgrave, 2016.
  • Crewwsell, Tim. In Place Out of Place: Geography, Ideology, and Transgression. Minneapolis: University of Minnesota Press, 1996.
  • Rhys, Jean. Quartet: Postures. London: Deutsch, 1969.
  • “What is open source?” opensource.com. Last modified 15 Oct 2015. http://opensource.com/resources/what-open-source.