Challenging new views on familiar
plotlines: A discussion of the use of
XML in the development of a
scholarly tool for literary pedagogy

............................................................................................................................................................

Monica Brown
Department of English, The University of British Columbia,
Vancouver, British Columbia, Canada
Teresa Dobson
Department of Language and Literacy Education, The University of
British Columbia, Vancouver, British Columbia, Canada
Dustin Grue
Department of English, The University of British Columbia,
Vancouver, British Columbia, Canada
Stan Ruecker
Institute of Design, Illinois Institute of Technology, Chicago, IL,
USA
.......................................................................................................................................

Abstract

Correspondence: Monica
Brown, Department of
English, The University of
British Columbia, 373-1873
East Mall, Vancouver, British
Columbia, Canada V6T 1Z1
Email: mm2brown@
interchange.ubc.ca

This article describes PlotVisML, a simple, flexible XML schema for encoding
literary narratives that was developed by an interdisciplinary team of researchers
in literary studies, interface design, computing studies, and education as part of a
research project on reading, writing, and teaching complex literary narrative.
PlotVisML is a simple, adaptable schema consisting of five key elements:
<action>, <dialogue>, and <narration> (tags for marking up narrative events),
and <character> and <object> (tags for encoding narrative objects). Fictional
narratives that have been marked up using PlotVisML can be visualized in
PlotVis, a digital scholarly tool that allows users to model and interact with
literary narratives in three dimensions. Both PlotVis, an interactive visualization
tool, and PlotVisML, our custom XML schema for encoding literary narratives,
were designed to permit challenging new views on familiar plotlines and, more
importantly, to depart from conventional ways of modeling narrative in
literary instruction. In discussing the process of developing PlotVisML, we
contribute to the ongoing discussion of text encoding as a form of close reading
(e.g., Liepert, 2009).

.................................................................................................................................................................................

Literary and Linguistic Computing, Vol. 28, No. 2, 2013. ß The Author 2013. Published by Oxford University Press on
behalf of ALLC. All rights reserved. For Permissions, please email: journals.permissions@oup.com
doi:10.1093/llc/fqt016
Advance Access published on 27 March 2013

199

M. Brown et al.

1 Introduction
This article discusses the use of Extensible Markup
Language (XML) to encode fiction by an interdisciplinary team of researchers in literary studies,
interface design, computing studies, and education
as part of a larger research project on reading, writing, and teaching literary narrative at the secondary
and postsecondary levels. One goal of the project
was to generate a digital scholarly resource that
would allow users to model and interact with literary narratives in three dimensions (see Fig. 1), thus
permitting different, unique perspectives on fictional plots and, more importantly, departing
from ways of modeling narrative that have traditionally relied on the Cartesian graph, such as
those described by Sterne (1847), Freytag (1863/
1983), and Vonnegut (1973), wherein time is
plotted on the x axis and the fortunes of the hero
or the disposition of the action (whether it is ‘rising’
or ‘falling’) is plotted on the y axis.
Elsewhere (Dobson et al., 2011), we discuss the
theory and design of our pilot model, PlotVis, an
interactive scholarly tool that visualizes XMLencoded narratives that can then be ‘customized
by the teachers and students in order to accommodate various interpretations of a single piece of fiction’ (p. 170). In that article, we also review
different models of plot visualization, and explore
the theory, or theories, of narrative implicit in each
one. For example, as we discuss there, in contrast to
interpretations of fiction prescribed by Freytag’s
static model, the perspective of which keeps readers’
focus primarily on the linear development of narrative action, PlotVis displays stories semantically
encoded in XML and includes features that allow
users to select, manipulate, highlight, and rearrange
significant features of a story, encoded as XML
elements that represented characters, objects, and
various narrative events. Users may visualize
stories in one of five designs, which emphasize different features of narrative, such as sequence and
structural complexity, and which allow users to explore and examine the myriad, multifaceted relationships between the narrative elements that
comprise a work of fiction (e.g. between character
and action).
200

Literary and Linguistic Computing, Vol. 28, No. 2, 2013

In this article, we describe the development of
PlotVisML, a custom XML schema that can be
used to encode fictional narratives so that they
may be visualized in PlotVis. We also give examples
of PlotVisML taken from an encoded version of the
title story of Munro’s (1998) The Love of a Good
Woman. PlotVisML was designed to provide users
of PlotVis, many of whom may be new to XML,
with a simple, adaptable schema for encoding fiction that could be applied to a wide range of fictional texts and that marks up five key elements of
narrative: action, character, dialogue, narration, and
objects. The schema was also developed to test pilot
models of the scholarly tool; present iterations of
PlotVis, however, allow users to visualize stories
encoded using custom XML schemas. Narrative features marked up as XML elements, whether using
PlotVisML or a user-designed schema, are then represented, in PlotVis, as the features, or ‘building
blocks’, of one of five different possible structures.
The goal of this article is to discuss the experiences of and challenges faced when using XML to
encode fiction, and to consider the implications of
these insights for future work in digital humanities
(DH) involving digital encoding of fiction and, further, for current literary pedagogy.

2 Coming to Terms With
Encoding Fiction in XML
Encoding fictional texts in this project involved
coming to terms with the idiosyncrasies of literature
and literary analysis and thinking about semantic
encoding as part of the process of literary interpretation—even when, at times, the rules of semantic
encoding seemed at odds with habitual methods of
closely reading a story. Fiormonte et al. (2010) observe, in their discussion of the use of XML in
encoding autographic literary texts, that ‘any transformation from paper to bits, apart from leading to
certain developments, all potentially ‘‘intrusive’’ for
the document, is no ordinary event, nor is it technically neutral’. The authors add, ‘in the moment
in which a text is transcribed through the selection
and use of markup this creates meaning itself’. We,
too, found that the encoding process generated

Challenging new views

Fig. 1 PlotVis (Dobson et al., 2010a).

meaningful observations about the relationships between textual markup and literary analysis comparable with those cited by Liepert (2009) and Flanders
(2011), among many others.
As noted previously, we focused our encoding
efforts on the title story from Munro’s (1998) The
Love of a Good Woman, the results of which were to
be visualized in the first iteration of PlotVis. We
selected this story, 89 pages long and >25,000
words, because it is demonstrative of Munro’s
theory of narrative as having a ‘soul’ around
which all other elements are layered (Ross, 2002),
a view that informs the design of one of our visualization prototypes (Dobson et al., 2010b), which,
based on a Fibonacci series, allows users to organize
all encoded elements of a story around one central
element, as selected by them. Using PlotVis’s ‘sort’
function, users can also arrange elements so that
their relation to the central element reflects either
proximity in time or else proximity in the story
(Dobson et al., 2010b, p. 4). That Munro’s theory
of narrative was influential to the PlotVis prototype
made it a relevant text to use in the development of
PlotVisML (see, e.g., Fig. 3).
As we developed PlotVis, we used Munro’s text
as a case study to determine which aspects of a
literary narrative might be represented as elements
in an XML schema, to consider how such elements
might relate to one another at the markup stage,
and to discuss the implications, both for literary

pedagogy and digital visualization, of using semantic encoding to highlight these features as significant. In other words, PlotVisML is not the
only XML schema that can be used to visualize
literary narratives in PlotVis, nor is it intended
to showcase the many possibilities for representing
narrative structure using textual markup. Rather,
PlotVisML was the result of our efforts to create
a simple, easy-to-use resource for users of PlotVis
who may be new to semantic encoding. Coming up
with a schema for encoding literary texts to
be visualized in PlotVis was also fundamental to
testing our three-dimensional (3D) visualization
system—it was part of the process of refining the
design of the interface and determining how it
might be used in literary analysis and literary
instruction.1
Establishing which aspects of narrative would
become the elements of PlotVisML was a process
of trial and error—of looking for ways to encode
story data in ways that were meaningful in terms of
literary criticism and that would facilitate valid
XML. A review of existing scholarship at the time
we began the encoding stage of our project, in 2010,
yielded only two other projects using XML to
encode fictional narratives: Proppian Fairy Tale
Markup Language (PftML; Malec, 2001; Lendvai
et al., 2010) and StoryML (Hu, 2003). PftML is
used in encoding fairy and folk tales, and its tags
mark up the 31 generic narratemes, or narrative

Literary and Linguistic Computing, Vol. 28, No. 2, 2013

201

M. Brown et al.

functions, identified by formalist scholar Vladimir
Propp in his extensive study of Russian folk tales
(Propp, 1968). StoryML, which was designed for
scripting interactive narratives to be presented in a
digital environment, points to the role XML can
play in the writing of complex fiction. Such projects
highlight the range of methods and applications for
semantic encoding of fictional texts.
More than fourteen versions of Munro’s text
were encoded, section by section, by two graduate
research assistant members of the research team
over a period of 6 months (the story is composed
of four long sections and an introduction). Both
encoders had strong backgrounds in literary criticism and literary instruction, yet, as is the case in
many funded DH projects that employ student researchers, both had minimal experience with semantic encoding at the time they began the
project. These researchers were first trained in different methods of textual markup and semantic
encoding and then instructed to use their knowledge
of literary criticism to determine which aspects of
Munro’s story should be marked up and how. The
process of development of PlotVisML, then, was inductive and iterative; once each section of the text
had been encoded in this way by both researchers,
separately, the project team met to review and compare the encoded sections of Munro’s text and determine which tags developed by the encoders
would be included in PlotVisML and why. Such
decisions were made on the basis of the following
criteria: similarities between encoder’s textual
markup (i.e. both encoders chose to mark up characters, and both also chose to use tags to distinguish
between narrative action and passages of narration);
the frequency of use of certain tags; and the implications of the tags as perceived by the team for
facilitating visualization in a three-dimensional
model.

3 PlotVisML Elements and
Attributes
PlotVisML is a simple, adaptable schema consisting
of five tags: <action>, <dialogue>, and <narration>, tags for marking up narrative events; and
202

Literary and Linguistic Computing, Vol. 28, No. 2, 2013

<character> and <object>, tags for encoding narrative agents. Fig. 2 is a diagram of the five narrative
elements that make up PlotVisML, as well as attributes that can be used, in encoding a literary narrative, to represent additional (and sometimes
implicit) story information—for example, to specify
which character is the speaker of the content of the
element, <dialogue>, or to indicate the story location of a narrative action.
PlotVisML is meant to structure, store, and
transport narrative information and, consonant
with the goals of XML, emphasizes simplicity, generality, and usability in encoding literary narratives.
Although the five elements that make up the schema
are flat and can therefore be nested within one another in many different ways, in encoded versions of
‘The Love of a Good Woman’, encoders frequently
found that narrative-agent tags, <character> and
<object>, typically branched out from the three
narrative-event tags, <action>, <dialogue>, and
<narration>, as in the following examples:
<action agent¼‘‘Enid’’ patient¼‘‘notebook’’
type¼‘‘record’’
location¼‘‘insideQuinn
House’’
keywords¼‘‘glomerulonephritis,
MrsQuinn’’>‘‘GLOMERULONEPHRITIS,’’
<character
reg¼‘‘Enid’’>Enid</character>
wrote in <character reg¼‘‘Enid’’>her </character><object reg¼‘‘noteBook1’’>notebook</
object>.</action>
<dialogue speaker¼‘‘MrsGreen’’ location¼
‘‘insideQuinnHouse’’
keywords¼‘‘death,ill
ness’’>‘‘How do <character reg¼‘‘you’’>you
</character> contract that kind of a disease
anyhow?’’
said
<character
reg¼‘‘Mrs
Quinn’’>Mrs. Quinn’s </character><character reg¼‘‘MrsGreen’’>sister-in-law</character>. <character reg¼‘‘MrsGreen’’>Her</
character> name was <character reg¼
‘‘MrsGreen’’>Mrs. Green</character>. <character reg¼‘‘MrsGreen’’>Olive Green</character>.</dialogue>
<narration narrator¼‘‘Munro’’ reporter¼
‘‘Munro’’ type¼‘‘description’’ location¼
‘‘FernsHouse’’ keywords¼‘‘CeceFerns,domestic,female,food’’>When the heat was more to
<character
reg¼‘‘CeceFerns’’>his</

Fig. 2 PlotVisML elements and attributes.

Challenging new views

character> liking, <character reg¼‘‘Cece
Ferns’’>he</character> put the <object
class¼‘‘tool’’ type¼‘‘domestic’’ keywords¼
‘‘care,food,danger’’>pan</object> down and
coaxed the <object class¼‘‘food’’ type¼‘‘eggs’’
keywords¼‘‘bodypart,female,reproduction,
violence’’>lacy edges of the eggs</object>
into
tidy
circles.
<character
reg¼
‘‘CeceFerns’’>He</character>
found
a
<object class¼‘‘tool’’ type¼‘‘domestic’’ keywords¼‘‘care,illness’’>clean spoon</object>
and dribbled a little <object class¼‘‘food’’
type¼‘‘fresh’’ keywords¼‘‘heat’’>hot fat</
object> over the <object class¼‘‘food’’ type‘‘eggs’’
keywords¼‘‘bodypart,female,reproduction,violence’’>yolks</object> to set
them. <character reg¼‘‘CeceFerns’’>He</
character> and <character reg¼‘‘Cece
Ferns’’>his</character><character
reg¼‘‘MrsFerns’’> mother</character> liked
<character reg¼‘‘CeceFerns,MrsFerns’’>their
</character> <object class¼‘‘food’’ type‘‘eggs’’
keywords¼‘‘bodypart,female,reproduction,violence’’>eggs</object> cooked this
way, but <character reg¼‘‘CeceFerns’’>his</
character><character
reg¼‘‘MrsFerns’’>
mother</character> often couldn’t manage
it right. <character reg¼‘‘CeceFerns’’>His</
character><character
reg¼‘‘MrFerns’’>
father</character> liked <character reg¼
‘‘MrFerns’’>his</character><object
class¼‘‘food’’
type¼‘‘eggs’’
keywords¼
‘‘bodypart,female,reproduction,violence’’>
eggs</object> turned over and flattened out
like
<object
class¼‘‘food’’>pancakes</
object>, cooked hard as <object class¼
‘‘food,clothing’’ type¼‘‘eggs,male’’>shoe leather</object> and blackened with pepper.
<character reg¼‘‘CeceFerns’’>Cece</character> could cook <object class¼‘‘food’’ type‘‘eggs’’
keywords¼‘‘bodypart,female,reproduction,violence’’>them</object> the way
<character reg¼‘‘MrFerns’’>he</character>
wanted, too.</narration>
Attending to nesting rules in XML therefore often
drew encoders’ attention to patterns in the relationships between narrative elements, reinforcing our
Literary and Linguistic Computing, Vol. 28, No. 2, 2013

203

M. Brown et al.

perception of semantic encoding as comparable
with the act of closely reading a text.2
Although observing nesting rules heightened our
team’s awareness of associations between elements
that comprise literary narratives, the process of developing a custom XML schema for encoding fiction
also raised questions about the influence of hierarchical arrangement on a reader’s understanding
and interpretation of a work of fiction. Team researchers from literature, especially, struggled in
the early stages with the potential implications, for
literary analysis, of the hierarchical relationship between elements in XML documents. Fiormonte et al.
(2010), similarly, note that the ‘‘well-formedness’’ of
XML, the requirement that every element, except for
the document root, must be contained within some
other element, gives the text an explicit and unambiguous structure’, a principle of this method of
semantic encoding ‘in contrast with what Buzetti
(2002) defines as the ‘‘dynamic instability of literary
texts’’’. Our study of secondary students’
approaches to XML encoding of literary narratives
(Grue et al., in press) revealed that, ‘Although in
principle, XML tags are binary and hierarchical, in
practice they are not’. Students more familiar with
literary criticism than semantic encoding were often
unable to produce valid XML in their experiments;
however, their inability to follow this rule often revealed serious engagement with the question of how
different elements of narrative may constitute a
text’s meanings. The insight we drew from that
study and our experiences in developing
PlotVisML led us to see these experiences as presenting new opportunities for critical inquiry.
For example, the perception, among some team
researchers, that XML encoding imposed or
overemphasized a hierarchical relationship between
narrative elements eventually led to an altered view
of the process of close reading. By marking certain
elements of a story as ‘significant’, such as the
protagonist (the perceived main character of a
story) or the climax (the most important event, as
suggested by Freytag), close reading similarly asserts
narrative to be constituted by a hierarchy, rather
than, perhaps, an ecology of agents and events.
The use of traditional close reading to raise certain
elements to a new level of distinction is in fact held
204

Literary and Linguistic Computing, Vol. 28, No. 2, 2013

in check in the act of marking up a story in
PlotVisML, and semantic encoding generally,
which requires that all narrative agents and events
be encoded, not only those that appear most
important to the critic based on customary models
such as Freytag’s Pyramid. XML schemas such as
PlotVisML can thus be used to test how meaning
in fictional narrative is produced—to figure
out, for example, what is meant by evaluations of
a fictional narrative as ‘character-driven’ or
‘action-driven’.3
Indeed, approaching literary text encoding as an
act of close reading and a method of theorizing narrative led to strenuous questioning by team members of the level at which meaning was made in
fiction: could a word in a story constitute a reader’s
idea of the significance of an object to the plot as a
whole? If not, how much additional information
could be supplied by way of attributes? How
might relationships between elements be expressed?
What are these relationships? Ultimately, it was
decided that although the design of PlotVisML
does suggest a theory of what constitutes narrative
fiction, these fundamental questions of meaning in
literature might be better inferred from its application to text rather than prescribed by its structure—
if PlotVisML’s purpose is to facilitate visualization
in PlotVis and critique more traditional visualizations of narrative relying on entrenched assumptions, we might question our own assumptions as
to not overly constrain its application. This influenced the development of PlotVisML as a minimal,
rather than complex, schema that would produce
valid XML and complex renderings of semantically
encoded texts but also allow for some degree of
freedom in its use.
Further, as earlier mentioned, texts encoded
using the five flat tags that comprise PlotVisML
are represented, in the PlotVis interface, as
the ‘building blocks’ of a story that can be rearranged into different structures and viewed from
multiple perspectives. The interface also facilitates
the examination of narrative elements in relation
to one another, as in the study of ecology: users
can, for example, highlight an element, such as
‘character’, and see how often this element is associated with ‘dialogue’ in the encoded text. Thus,

Challenging new views

Fig. 3 Displaying elements and attributes in PlotVis.

even texts that have been encoded, or analyzed, in a
manner that emphasizes hierarchy can be explored,
in PlotVis, apart from the connotations such associations may imply at the level of encoding.
Attributes are provisional to PlotVisML, but may
become more important in the resultant visualization, depending on whether text encoders will also
be using the PlotVis interface to analyze and interpret
this text. PlotVis displays each encoded narrative
element individually; users can mouse over an element and read the value of its attributes. The information provided in attributes therefore helps users
situate unique elements both in the context of the
original story and within the new view permitted by
the 3D visualization (see Fig. 3).
For example, ‘reg’, an attribute of the
<character> tag, can be used to supply the proper
name of a character referenced within a story, a
function that is especially helpful in fictional narratives, which frequently contain pronominal references, as in the following case, in which the
attribute ‘reg’ makes it possible to assign to the

pronoun, ‘they’, the names of three separate characters being discussed by the narrator:
At least, <character reg¼‘‘CeceFerns,Bud
Salter,JimmyBox’’>they</character> would
call it swimming; <character reg¼‘‘Cece
Ferns,BudSalter,JimmyBox’’>they</character> would go back to town and say that
<character reg¼‘‘CeceFerns,BudSalter,Jimmy
Box’’>they</character> had been swimming
at Jutland before the snow was off the ground.
Furthermore, attributes allow for the association
of the narrative-event tags—<action>, <narration>,
and <dialogue>—with characters directly involved
in these events. For example, <action> can be associated with characters using the attributes ‘agent’
and ‘patient’, as can <dialogue> using the attribute
‘speaker’:4
<action agent¼‘‘Enid’’ patient¼‘‘Lois,Mrs
Quinn,Sylvie’’ type¼‘‘demonstration,instruction’’ location¼‘‘outsideQuinnHouse,inside

Literary and Linguistic Computing, Vol. 28, No. 2, 2013

205

M. Brown et al.

QuinnHouse’’ keywords¼‘‘female,girls,playthings,MrsQuinn’’>On a Saturday morning,
<character
reg¼‘‘Enid’’>Enid</character>
called <character reg¼‘‘Louis’’>Lois</character> and <character reg¼‘‘Sylvie’’>Sylvie</
character> from <character reg¼‘‘Louise,
Sylvie’’>their</character> games under the
porch, to come and see <character
reg¼‘‘Louise,Sylvie’’>their
</character>
<character reg¼‘‘MrsQuinn’’>mother</character> looking pretty.</action>
dialogue speaker¼‘‘MrsQuinn’’ location¼
‘‘insideQuinnHouse’’
keywords¼‘‘anger,
domestic,Enid,girls’’><character reg¼‘‘Mrs
Quinn’’>Mrs.
Quinn</character>
said,
‘‘Keep <character
reg¼‘‘Louise,Sylvie’’>
them</character>
off
of
<character
reg¼‘‘MrsQuinn’’>my</character>
bed,
<character
reg¼‘‘Louise,Sylvie’’>they’re</
character> filthy."</dialogue>
The event-tag, <narration>, includes the attributes ‘narrator’ and ‘reporter’, which make it
possible to distinguish, through encoding, between different levels and styles of narrative points
of view:
<narration
narrator¼‘‘Enid’’
reporter¼
‘‘Munro’’ type¼‘‘record’’ location¼‘‘inside
QuinnHouse’’ keywords¼‘‘death,food,illness,
MrsQuinn,notebook’’>‘‘JULY. Rain early
a.m. <character reg¼‘‘Louis’’>L.</character>
and <character reg¼‘‘Sylvie’’>S.</character>
playing under porch. <object class¼‘‘appliance’’ type¼‘‘fan’’ keywords¼‘‘wind’’>Fan</
object> off and on, complains noise. Half
cup
<object
class¼‘‘food,drink’’
keywords¼‘‘egg,illness,childhood,care,healthcare,
profession,female’’>eggnog</object><object
class¼‘‘cutlery’’> spoon</object> at a time.
B.P. up, pulse rapid, no complaints pain.
Rain didn’t cool off much. R.Q. in evening.
Hay finished.</narration>
In this example, the attribute ‘narrator’ is used to
assign a passage of narration to a character, Enid,
from whose perspective that passage is being told.
The attribute, ‘reporter’, is used to indicate another
206

Literary and Linguistic Computing, Vol. 28, No. 2, 2013

level of narration; in this case, it distinguishes the
author, Alice Munro, as the reporter of the narration itself. The distinction between narrator and reporter, although optional in PlotVisML, allows the
encoder to account for the layers of narration that
often comprise the effect of more experimental literary narratives, such as ‘The Love of a Good
Woman’.
Thus, in some instances, attributes serve a practical purpose in PlotVisML: they make it possible
for encoders to include information that could not
be inferred by users of PlotVis, as they interact with
story elements within the interface. However, attributes can also be used to provide metadata that may
be relevant to literary analysis. Users interested in
exploring the layers of narration that make up
Munro’s story can, for example, use PlotVis’s
search functions to isolate ‘narration’ elements according to attribute metadata.

4 Text Analysis, Literary Analysis
Digital humanities scholars have examined the role,
purpose, and implications of the computer in textual analysis (e.g., Rockwell, 2003; Renear, 2004).
Existing scholarship models, and critiques, a range
of methodologies for engaging digital technology in
textual criticism. Some of this scholarship addresses
the applications of digital technology in literary analysis in particular. Projects such as Clement’s (2008)
distant reading of Gertrude Stein’s The Making of
Americans illustrate the use of textual encoding and
text-analysis tools to challenge and augment
long-standing interpretations of canonical works
of fiction founded in conventional, print-based
methods of literary criticism. Yet, whether applied
in the analysis of canonical or emergent fictional
texts, or of print-based or digital-born narratives,
digitally augmented methods of literary scholarship
unsettle the fundamental customs of the discipline.
Such methods emphasize the role and function of
the critic as much as of the computer screen, and
challenge literary critics to be aware of and transparent in regards to what they pay attention to when
reading and interpreting a literary text.5

Challenging new views

Developing a method for semantically encoding
texts for use in PlotVis—and, more generally,
experimenting with ways of translating aspects of literary criticism into digital media—afforded further
insight into the view of text encoding as a form of
close reading and encouraged us to adapt our understanding of the process of literary analysis. The iterative process of PlotVisML’s development contrasted
not only multiple versions of prospective schemas but
also multiple readings of text, motivating the desire for a design that expresses the subjectivity of
the encoder while keeping it accountable to the material of narrative fiction—as well as the relationships
therein. This process therefore relied on a recursive
strategy of design and use, essentially an experimental
probing of narrative to reveal its secrets. That this
process was productively iterative—sometimes the
encoding of narrative defied expectations of the
schema’s use—further motivates our inquiry into
new views on schema development and the text
they encode.
More information about PlotVis, including
a video demonstration of the 3D visualization
system, can be found at http://blogs.ubc.ca/plotviz/.

References
Buzetti, D. (2002). Digital representation and the text
model. New Literary History, 33: 61–88.
Chaturverdi, M., Gannod, G., Mandell, L.,
Armstrong, H., and Hodgson, E. (2012). Myopia: A
Visualization Tool in Support of Close Reading. Digital
Humanities 2012. Hamburg, Germany: University of
Hamburg. http://www.dh2012.uni-hamburg.de/conference/programme/abstracts/myopia-a-visualizationtool-in-support-of-close-reading/ (accessed 3 January
2013).
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: 361–81.
Dobson, T. M., Michura, P., Rodriguez, O., and
Ruecker, S. (2010a). PlotVis [Computer software].
http://blogs.ubc.ca/PlotViz (accessed 27 February
2010).
Dobson, T. M., Michura, P., and Ruecker, S. (2010b).
Visualizing plot in 3D. Proceedings of the Fourth

International Conference on Digital Society. St.
Maarten: Netherlands Antilles, February 10–16, 2010.
Dobson, T. M., Michura, P., Ruecker, S., Brown, M., and
Rodriguez, O. (2011). Interactive visualizations of plot
in fiction. Visible Language, 45: 169–91.
Fiormonte, D., Martiradonna, V., and Schmidt, D.
(2010). Digital encoding as a hermeneutic and semiotic
act: the case of Valerio Magrelli Digital Humanities
Quarterly, 4,1. http://www.digitalhumanities.org/dhq/
vol/4/1/000082/000082.html (accessed 3 January 2013).
Flanders, J. (2011). Possible Worlds: Authorial Markup
and Digital Scholarship Digital Humanities 2011. Palo
Alto, CA: Stanford University.
Freytag, G. (1863/1983). Die Technik Des Dramas. Leipzig:
S. Hirtzel.
Grue, D., Dobson, T., and Brown, M. (in press). Reading
practices and digital experiences: an investigation into
secondary students’ reading practices and XMLmarkup experiences of fiction. Literary and Linguistic
Computing, 28: 237–48.
Hemingway, E. (1991). A Clean, Well-Lighted Place. In
Carl, B., Beaty, J., and Hunter, J. P. (eds), The Norton
Introduction to Literature. New York: Norton,
pp. 393–7(Originally published in 1926).
Hu, J. (2003). StoryML: enabling distributed interfaces for
interactive media. Proceedings of the Twelfth
International World Wide Web Conference. Budapest:
Hungary.
´
Lendvai, P., Declerck, T., Daranyi, S., and Malec, S.
(2010). Propp Revisited: Intergration of Linguistic
Markup into Structured Content Descriptors of Tales.
Digital Humanities 2010. London, UK: Oxford
University Press.
Liepert, S. (2009). Richardson’s Clarissa and the future of
close reading. Presentation Given at the Digital Literacy
Centre, Faculty of Education, The University of British
Columbia. Vancouver, BC.
Malec, S. A. (2001). Proppian structural analysis and
XML modeling. http://clover.slavic.pitt.edu/sam/propp
/theory/propp.html (accessed 15 October 2010).
Munro, A. (1998). The Love of a Good Woman.
Hammond, UK: Penguin Books, pp. 1–89.
Propp, V. (1968). Morphology of the Folktale. Indiana:
Indiana University and the American Folklore Society.
Renear, A. H. (2004). Text Encoding. In Schreibman, S.,
Siemens, R., and Unsworth, J. (eds), A Companion to
Digital Humanities, ch. 17. http://www.digitalhumanities.org/companion/ (accessed 21 September 2011).

Literary and Linguistic Computing, Vol. 28, No. 2, 2013

207

M. Brown et al.

Rockwell, G. (2003). What is text analysis, really? Literary
and Linguistic Computing, 18: 209–19.
Rommel, T. (2004). Literary studies. In Schreibman, S.,
Siemens, R., and Unsworth, J. (eds), A Companion to
Digital Humanities, ch. 8. http://www.digitalhumanities.
org/companion/ (accessed 10 September 2011).
Ross, C. (2002). ‘Too many things’: Reading Alice
Munro’s ‘The love of a good woman’. University of
Toronto Quarterly, 71: 786–810.
Sterne, L. (1847). The Works of Laurence Sterne,
Containing the Life and Opinions of Tristram Shandy,
Gentleman. London: J. J. Chidley.
Vonnegut, K. (1973). Breakfast of Champions. London:
Cape.

2

3

Notes
1 Earlier versions of this article discussed the rationale for
PlotVisML with reference to the TEI’s P5 Guidelines for
Electronic Text Encoding and Interchange, which was
an important resource used by our research team, as we
sought a method for encoding fictional text to be visualized in PlotVis. Indeed, TEI was an option we considered in the early stages of this project, and the
decision to create a custom XML schema to use with
PlotVis was made not on the basis of perceived shortcomings of using TEI so much as on an interest in
coming up with tags that would be familiar to novice
literary critics (e.g. students at the secondary and postsecondary levels) who would be using PlotVis, many of
whom, we expected, would have little to no previous
experience of semantic encoding. We also wanted tags
to reflect, as much as possible, the concepts expert literary critics use to guide their analyses of fictional texts.
In this way, our visualization tool is not unlike
University of Miami’s poetry visualization tool,
Myopia (Chaturverdi et al., 2012), in which, ‘Even
though one schema is used, a schema extended from
the TEI schema, the elements of a poem that are interesting to literary scholars have been formalized as tags
and attributes based on XML standards’. Adopting
XML does not rule out the future use of TEI in this
project. As one reviewer pointed out to us, ‘For data of
this kind (with such specific applications and such a
small/task-specific tag set), TEI seems useful chiefly as

208

Literary and Linguistic Computing, Vol. 28, No. 2, 2013

4

5

an interchange format, and it would be perfectly
straightforward to export/convert [this] data to TEI if
necessary for sharing or archiving’. That is, although
texts encoded using PlotVisML can be exported to
TEI if necessary, the choice to use a simple, custom
XML schema was a matter of practicality: creating
custom tags was less time-consuming than creating a
TEI customization and, more importantly, has resulted
in a tag set that reflects the interests of literary critics
and that may be used by those with little or no previous
semantic encoding experience.
Once encoded texts have been visualized in PlotVis,
relationships between narrative elements noticed at
the encoding stage can be further examined.
The possibility of representing the five narrative elements in a non-hierarchical encoding structure, such as
the Resource Description Framework data model, was
also discussed as a potentially more flexible method of
encoding the overlapping and often asymmetrical relationships between narrative elements. This approach is
presently being explored by our research team. We have
also experimented with uses of PlotVisML that incorporate ‘Ographies’, such as personographies, placeographies, and, due to the significance of objects in
narrative fiction, objectographies. These structures
offer different methods of associating narrative elements and including thematic and interpretative information in an encoded text.
It has been suggested that, as the element, <action>,
carries attributes for ‘agent’ and ‘patient’, the attribute
<dialogue> might similarly carry both ‘speaker’ and
‘auditor’. Not only can PlotVis easily accommodate
changes and additions to PlotVisML, the attribute ‘auditor’ is a perfect example of an addition that could be
made to make the schema suit the story—for instance, in
a short story such as Hemingway’s (1926/1991) ‘A Clean,
Well-Lighted Place’, in which long passages of dialogue
may make it difficult for readers to keep track of who is
speaking and who listening.
Rommel (2004), for example, describes literary computing as a method of substantiating literary interpretations by making possible ‘the identification of strings
and patterns in electronic [or digitized] texts’, thereby
legitimating the subjectivity of the literary critic: in
other words, confirming literary interpretation to be a
disciplinary and, to an extent, empirical process.

Copyright of Literary & Linguistic Computing is the property of Oxford University Press / USA and its content
may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express
written permission. However, users may print, download, or email articles for individual use.