The 3DH project aims to lay the foundations for a ‘next-generation’ approach to visualisation in and for the Humanities. As for the theoretical background, the project’s frame of reference are the particular epistemological principles that are relevant to hermeneutic disciplines and which must therefore also orientate our approach to visualisation.

For the 3DH visualisation concept we formulate four postulates. These are

1. the “2 way screen postulate” (i.e. an interaction focused approach toward visualisation);
2. the “parallax postulate“ (i.e. the idea that visualisation in and for the humanities should not just tolerate, but actively put to use the power of visual multiperspectivity in order to realise epistemic multiperspectivity);
3. the “qualitative postulate” (i.e. the idea that visualisations should not just ‘represent’ data, but also offer a means to make and exchange qualitative statements about data);
4. the “discursive postulate” (i.e. the idea that visualisations should not just be used to illustrate an already formed argument or line of reasoning, but should also become functional during the preceding/subsequent steps of reasoning, such as exploration of phenomena and data, generation of hypotheses, critique and validation, etc.).

During the 2016 summer term we organized a public lecture series on DH visualisations (see also 3DH blog and  https://lecture2go.uni-hamburg.de/l2go/-/get/v/19218).

One outcome of the lecture series was a need for bringing in the expertise of visual design specialists. By bringing together the “two worlds” of literary studies and visual design we hope to transcend the limitations of our respective visual(ising) routines.

Co-teaching seminar University of Applied Science of Potsdam and University of Hamburg

In the 2016 summer term we have begun to engage in a co-teaching project with the visual design specialists Marian Dörk and Jan Erik Stange from the University of Applied Science Potsdam. Two groups of students meet during four workshops alternately held in Potsdam and in Hamburg, one a class of German literature master students (Prof. Chris Meister, Universität Hamburg), the second a class of design students (Prof. Marian Dörk).  Their joint goal is  to answer two questions:

• ‘In how far can visualisations be helpful for the analysis of literary texts?’ and
• ‘Where do visualisations have their place in a subjective and interpretive structure?’

The literary text under discussion is the novel Johannisnacht by the German author Uwe Timm, published in 1996. It tells the story of a writer suffering from writer’s block, who gets the opportunity to write a report about the history of the potato. As trivial as this task seems to be at first, the writer’s research becomes more and more a life-threatening and odd adventure. In a formal aesthetic interplay between narratological categories such as the narrator, the discursive parameters of time and place and metafictional elements of self-reflection, Johannisnacht stretches out the genesis and the specific functionalities of narration.

Our two groups approach this text from two different perspectives: the German literature students try to identify and define possible visualisation needs for their work in text analysis (in which they focus on some of the novel’s narratologically salient aspects). The task of the design students on the other hand is to consider the literary text as a whole and find practical solutions for visualising those of its structural features that might be pertinent to the literary scholar’s analytical needs. The combination of both approaches – the narratological/literary studies perspective with the visualisation design perspective- is our first step toward defining a specific narratological visualisation use-case.

First Co-Creation Workshop and its results

After a short input presentation on relevant narratological concepts and methods (Chris Meister) we tried to gain a first understanding of some of the novel’s structural features and then co-created first drafts of visualisations. Our sketches focused on three questions:

• What narratological questions are raised by reading and analysing the whole novel or a single chapter of Johannisnacht?
• What texts and data are required for answering this question?
• What kind of visual representation could be suitable for answering this question?

Here are our first visualisation ideas:

This sketch tries to point out the narratological category of focalisation in the course of the story.

This sketch shows an attempt of structuring the story of the novel by describing the relationship between character, narrator and narrated objects.

This group gains an access to Johannisnacht by asking ‘What generates complexity in a novel?’

This sketch shows the attempt to visualise the narrative levels of Johannisnacht.

This group focuses on the character constellation in the novel Johannisnacht: Who is talking about whom? On which narrative level is a character introduced?

Merken

The poetry scholar Katherine Coles gave a presentation on Poemage at the VIS4DH workshop at VIS2016. Poemage is a tool for exploring and analysing the sound topology of a poem. It is an interdisciplinary work between poetry scholars, computer scientists and linguists. Recommended reading is not only the presented paper Show ambiguity, which takes a more poetry scholar influenced perspective on Poemage but also the companion paper which complements “Show ambiguity” by adding the computer scientist stance to it. Besides the methodological principles that are covered by Poemage both papers give also great insight into the collaborative aspects of the project across disciplines.

The UI of Poemage offers three views. The Set View offers rhyme sets, which are sets of words that are connected by a specific rhyme scheme. The rhyme sets are organized by rhyme types. Each circle represents a specific rhyme set. The size of the circle depends on the number of words in the set. The Poem View shows the poem in its original form and the Path View gives a 2D space where the flow of the poem according to its rhyme topology is displayed. Each node in the path view represents a word in the poem and is positioned in relation to its position in the layout of the poem. The curves show the flow of a rhyme set through the poem. The views are linked by color coding and by interaction: e. g. selecting a rhyme set in the Set View also activates the visualization of that rhyme set in the other two views.

I like especially the openness of the tool. It supports and encourages multiple readings and the rhyme types are extensible in two ways. The simple way allows the scholar to group words freely to form custom sets without being bound to any predefined rhyme type. The more complex way allows the scholar to access the underlying rules engine or formalism to formulate new rhyme types in a notation which is geared to poetry scholars.

The representation of rhyme sets as paths allows exploration of the rhyme topology by examining spatial phenomena of the paths like intersections, mergings and divisions. There is a tight link between the visualisation and the poem that makes it easy to trace back observations in the visualization to the original data.

Another interesting aspect of her talk was when Coles shared her view on the humanistic idiosyncrasies of data visualization, especially in poetry scholarship. She wanted Poemage “to provide an aesthetically enriched experience” and emphasized the engagement between scholar and object of study which should extend to the visualization as well.

When we discussed the special needs for the humanities for visualization in the 3DH project so far, I (with a computer science background) was very sceptical about seeing the humanities on one side and the hard sciences on the other side. On the contrary I can see a lot of common ground between a physicist and a humanities scholar exploring and interpreting his or her data with visualizations. Instead of seeing the two as opposites we in 3DH started to work with a methodological continuum between the poles of subjectivity/uniqueness/particularity and objectivity/reproducibility/universality. I doubt that the kind of engagement Coles describes is the same engagement between a physicist and his or her data. I think Coles managed to describe at least part of the possible contribution of visualisation to one extreme of that continuum. And this really helps to track down the methodological differences 3DH visualizations need to account for.

Some of the things we have to think about when reading historical visualizations include:

• Classification – how is the world classified when the visualization was drawn up?
• Derived vs manually produced data – how did the data get to the cartographer and, for that matter, how did the map get to us?
• Graphic vs. textual representations – we are continually transforming representations from visual to textual and back – what happens in the transcoding?
• Epistemology – how do we know what we think we know?
• Time and change – how is time and change collapsed in representations of space?
• Completeness – we never have complete information, but sometimes we think we do
• Data proxies – we are not interacting with the phenomenon itself, but with surrogates
• Geography – what is special about the world?

He then showed 4 case studies.

Case Study 1: Roman Itineraries

Roman “station lists” exist that tell us about the stations if you went from somewhere to somewhere. There are a number of these lists, but we don’t know why they were created or stored. He showed an image of the Vicarello Goblets.

Then he showed ways to put these itineraries on a map or to visualize them in other ways like a topological map. A topological map can show closeness or betweeness. High betweeness is where you might get a bottleneck of travelers.

He showed that the data for itineraries has been put into databases for others to look at like that of the Ancient World Mapping Center. When we look at them the beginnings and ends of itineraries are often strange.

Leif commented that we need to be careful when drawing conclusions from small sets of data (like these lists) or combined data from different sets (and lists). We need to remember that the texts or lists are not the phenomenon. Access to texts is also changing rapidly.

Case Study 2: Ptolemy’s Geography

Ptolemy was famous early on for astronomy and geography. His Geographia was one of the earliest tools for geography. Ptolemy wrote about projections and provided a catalogue of places with coordinates that could be projected. We don’t have his maps, though we have maps created from his Geographia, to get those we try to project them from his theory and coordinates. (See Leif’s paper on Lines, damned lines and statistics: unearthing structure in Ptolemy’s Geographia (PDF).)

Leif showed a graph of the coordinate values that represented points in terms of time. Latitude is tied to length of longest day, but it isn’t even. You get an uneven grid. The coordinates of boundary locations seem to fall on the irregular time-based grid network. It is as if the edges of regions are meant to fall on the hours of the degree based system. It looks as if alignment was important to him – he seemed to have believed that alignment should be the case like grain in wood.

Leif talked about how Ptolemy’s data can tell us things about his theory. He talked about the warp and weft of the data. It is also important to see how visual arguments help us understand numeric ones at scale. The visual projection shows us something about Ptolemy’s coordinates (numbers).

Finally, Ptolemy is interested in mapping the world to the celestial which means he has to know the time.

Case Study 3: The Peutinger Map

The Peutinger map was his third case study. It is a copy of a map that we are fairly sure was produced in late antiquity. It is a weird parchment scroll that is long, but not high. Is it a guide for carrying on a trip? Rome is somewhere in the middle but not exactly in the middle which has led people like Richard Talbert in Rome’s World to argue that the long map was a presentation piece and Rome should be in the middle which means we are missing three parchments on one side.

Leif runs a project called Pelagios Commons which allows him to compare locations in the map to those from other itineraries (lists). The itineraries on the coast lines of the map seem similar to another work. He argued in support of Talbert suggesting that the scroll may have been copied from a wall.

This led to the general warning that the way that data is experienced and consumed is often very different now from how it was intended to be consumed and contextual evidence can come from other unexpected places.

Case Study 4: the Pelagios Map Tiles

Pelagios is aggregating information from early geographic documents. He showed how we need to be careful about flat maps – they can hide information. When you look at the 3D then you can begin to understand some of the flat decisions – like why you need to skirt certain mountains or wet areas (like the Po valley.)

Conclusions

Contemporary datasets are just like the old ones. They are composites, each with a long and complex heritage. They are incomplete and degraded. They need to be read closely and distantly. They need to be challenged just as we would any other type of evidence or claim.

He ended by saying that the point of the humanities is not to find answers so much as to question that answers we have been given. We need to rethink what we thought we knew and that includes visualizations.

On May 25th 2016 Johanna Drucker gave the second Susan Hockey Lecture in Digital Humanities at UCL to speak about the 3DH project and “Graphic Provocations: What do digital humanists want from visualization?”

Following the link you’ll find her lecture online: https://www.ucl.ac.uk/dh/events/SusanHockeyLecture/2016