The Theory and Design of PlotVis

The Theory and Design of PlotVis

Dobson, Teresa M., University of British Columbia, Canada, teresa.dobson@ubc.ca
Ruecker, Stan, The University of Alberta, sruecker@ualberta.ca
Brown, Monica, mm2brown@interchange.ubc.ca University of British Columbia, Canada,
Rodriguez, Omar, omar.rodrigueza@gmail.com The University of Alberta,
Michura, Piotr, zemichur@cyf-kr.edu.pl Academy of Fine Arts, Krakow, Poland,
Grue, Dustin, dgrue@shaw.ca University of British Columbia, Canada,

PlotVis is a 3D system for people interested in examining narrative action, or plot, from a variety of perspectives. The system was developed as one outcome of a funded study examining the teaching of complex narrative forms in secondary and undergraduate classrooms. Results of this study have revealed, inter alia, that conventional approaches to teaching fiction, particularly at the secondary level, fail to take account of the diversity of contemporary narrative (cf Dobson, 2006).

For example, instructors, particularly at the secondary level, still rely heavily on the five-stage plot mapping first described by Gustav Freytag (1863) in Die Technik des Dramas (Figure 1), and the superimposition of this model on forms beyond those it was originally intended to describe can be misleading (Dobson, Michura, Ruecker, 2010).

Figure 1: Freytag’s Pyramid shows five basic components of plot, based on Greek and Shakespearean tragedy.

Full Size Image

In addition, the reliance on the Cartesian graph to model narrative generally – and there are a number of examples of such graphing (e.g., Sterne, 1847; Vonnegut, 1973) – is restricting because it fails to take account of the multiple dimensions of story.
Considering the limitations of the Cartesian graph for modeling narrative along with conceptualizations of narrative as multidimensional (e.g., Shields, 2000), our goal was to produce a three-dimensional digital environment that reifies different perspectives on plot, so that students and other scholars can quickly shift from one 3D object to another, or spend time exploring any of the visualizations in more detail. We currently support three perspectives and are beginning to experiment with the design of a fourth.

This panel will examine the results of the PlotVis project from several different perspectives, including an introduction to the project, schema design for narrative encoding, a reading practices study, and design and programming.

Introduction to PlotVis as a Form of Distant Reading
… what we really need is a little pact with the devil: we know how to read texts, now let’s learn how not to read them. Distant reading: where distance, let me repeat it, is a condition of knowledge: it allows you to focus on units that are much smaller or much larger than the text: devices, themes, tropes—or genres and systems. (Moretti, 2000, p. 57).
Recently, literary theorists and digital humanists alike have taken up Moretti’s appeal, later developed in Graphs, Maps, Trees (2005), for the development of new methods of textual analysis. Clement (2008), for example, uses mapping techniques and visualizations to facilitate, quite literally, new perspectives on a canonical text, Gertrude Stein’s The Making of Americans. By premising knowledge on distance rather than closeness, Clement discerns significant patterns in the narrative’s structure, in so doing discovering a new logic to Stein’s text. Such new insights, made possible by adapting the scale and focus of textual analysis, are exactly what Moretti expects readers will earn by learning how not to read texts—and, by extension, by learning how to use different technologies to complement, and corroborate, our readings of texts.

Experiments with distant reading challenge the truism that readers make sense of a story by seizing upon a narrative episode, using it as a key to gain entry into the deeper meaning inherent within the text, “as if meaning resided in a buried treasure chest or behind a lock door” (Clement, 2008, p. 365). Clement proposes a different metaphor for understanding narrative structure, that of a map key, which instead implies that readers make sense of stories by using narrative structure as a guide. The metaphor of the map key offers a view of reading complex narrative as a process of orientation, through which readers manage their awareness of a text’s meanings by acquainting themselves with its structure.

Similar views of distant reading and complex narrative inform the project of developing XML schemas for encoding digitized versions of fictional narratives. Undertaken as part of a larger study based out of the University of British Columbia, this project investigates, and proposes, methods of reading adapted to contemporary print and digital fiction’s “shift away from conventional narrative logic toward indeterminacy, fragmentation, and open-endedness” (Dobson, 2008, p. 1). While the goal of our larger study is the development of new models for writing, reading, and teaching complex print and digital narrative, the focus here will be the use of XML to encode fictional narratives as, in itself, a form of textual analysis, involving both closeness and distance, and enabling even further alterations, of the kind Moretti proposes, to the scale and focus of scholarly reading practices.

Towards a Schema Design for Narrative Encoding
This paper will discuss the iterative development and use of a set of Extensible Markup Language (XML) schemas for encoding digitized versions of fictional narratives. Although the Text Encoding Initiative (TEI) Guidelines for Electronic Text Encoding and Interchange define and document a thorough markup language for encoding humanities texts, including significant provisions for encoding scholarly editions of drama, poetry, and manuscripts, there presently exist only a handful of XML schemas designed specifically for marking up literary fiction (e.g., StoryML, FicML, PftML).

Drawing on principles and approaches from narratology, our literary encoding schemas combine TEI elements and attributes with tags that specifically mark up elements of narrative structure, such as actions, characters, dialogue, narration, objects, places, and time. Examples of encoded short stories, such as Alice Munro’s “The Love of a Good Woman” and Ernest Hemingway’s “Hills Like White Elephants,” will be used to demonstrate, and critique, the different possibilities we have explored for developing literary XML schemas.

From Envisaging to Visualization: Young People’s Narrative Reading Practices
This paper will present findings of a study examining the reading practices of young adults. The study focused on the dialectic between how students of literature envisage narrative and how they actually visualize it in systematized approaches. Fifty participants in grades 11 and 12 read Hemingway’s “Hills Like White Elephants” and O’Faolain’s “The Trout.” They then engaged in a series of activities: evaluating form by physically “cutting and pasting” a narrative’s text, sketching narrative diagrams, XML tagging of narratives, and an exit interview.

Participants indicated that narratives were enjoyable because their plots were “unexpected,” but also un-enjoyable because they were “hard to understand.” That is, comprehension of a narrative is a condition of enjoyment and disappointment: narratives ought to be original but not confusing. In XML tagging, grade 11 students were given a basic schema to tag a narrative but were also encouraged to modify this schema. The schemas were altered beyond a systematized approach, instead revealing critical activity across a range of “levels.” Globally, for example, subjects modified the schemas beyond a “systematic” approach in order to highlight salient features particular to a text. Subjects added the tags setting and contradictory statements to the XML schema for “Hills,” suggesting a post-critical (and almost juridical) attitude. More locally, there is evidence that tagging is dependent on “spheres of lexical activity,” where tagging of features depends on the immediate context in which they occur. Furthermore, tagging appears to be not only independent of syntax, but also independent of the tagged-narrative’s grammar – subjects prescribe systematic rules, but do not follow them.

PlotVis: Design and Implementation of Five Plot Models
Our final panelists will discuss issues related to designing and programming four different views of narrative structure. In the first case, the user is able to see a Fibonacci series where one of the encoded plot elements is placed at the centre of the visualization, and other elements are arranged out from the centre, either sequentially as they appear in the text or else hierarchically as they are arranged in the XML tree (Figure 2).

Figure 2: This screenshot of Munro’s “Love of a Good Woman” demonstrates the complexity of the tagging, with each coloured cylinder representing one of roughly 7000 tagged pieces of text.

Full Size Image

In this visualization, each piece of text appears as a coloured cylinder. The height of the cylinders can be associated with the length of the text they contain. The second visualization privileges sequence over centrality, with multiple pipelines representing the characters or narrator, and changes to the arrangement of the pipe showing where changes in time occur (Figure 3).
Figure 3: Our design sketch for a view that privileges sequence over centrality, with multiple pipelines representing the characters or narrator.

Full Size Image

The third visualization is based on an architectural metaphor, where readers see the different kinds of encoded plot elements as floors of a building (Figure 4).
Figure 4: Our design sketch of the architectural concept of story visualization.

Full Size Image

Text that occurs only within a single element tag fills a space on a single floor, while text that spans multiple elements appears as blocks that span the floors. Our fourth and most recent design resembles in some ways our sequential model, only in this case the text is represented by vertical walls of text that bend at locations where there are changes in time, while the other encoded plot elements appear as colours on the surfaces of the walls (Figure 5).
Figure 5: This screenshot shows the wall display. Vertical coloured bands represent the encoded text elements that are shown as nodes in Figure 2.

Full Size Image

The visualization system was created using a popular game engine, called Unity3D, which provides us with a flexible platform for interaction with 3D objects, primarily through writing Javascript. The contents of the visualization, as indicated above, are produced from XML-encoded text files.
References:
Clement, T. E. 2008 “‘A thing not beginning and not ending’: using digital tools to distant-read gertrude stein’s the making of americans, ” Literary and Linguistic Computing, 23(3) 361-381 Dobson, T.M. 2008 Reading, Writing, and Teaching Complex Narrative, Standard Research Grant Application submitted to the Social Sciences and Humanities Research Council of Canada.,

Dobson, T. M. 2006 “For the love of a good narrative: Digitality and textuality, ” English Teaching: Practice and Critique, 5(2) 56-68

Dobson, T. M. Michura, P. Ruecker, S. 2010 “Visualizing plot in 3D, ” Proceedings of the Fourth International Conference on Digital Society,

Freytag, G. 1863/1983 Die technik des dramas,

Hemingway, E. 1927 Men Without Women,

Jessop, M. 2008 “Digital visualization as a scholarly activity, ” Literary and Linguistic Computing, 23(3) 281-293

McCarty, W. 2002 “Humanities computing: essential problems, experimental practice, ” Literary and Linguistic Computing , 17(1) 103-125

Moretti, F. 2005 Graphs, maps, trees: Abstract models for a literary theory,

Munro, A. 1998 The love of a good woman,

O'Faolain, S. 1980 “The trout, ” Collected stories. (Vol. I), 383-386

Shields, C. 2000 ““Ilk”, ” Dressing up for the carnival, 53-60

Sterne, L. 1847 The works of Laurence Sterne, containing the life and opinions of Tristram Shandy, gentleman,

Van Peer, W. Chatman, S. B. 2001 New perspectives on narrative perspective,

Vonnegut, K. 1973 Breakfast of champions,