View source for Using Theme to Author Hypertext Fiction

According to Prince (2003), a story can be seen as having three different types of macrostructures or frames: action (plot), existents (characters and setting), and ideas (theme). Research in interactive storytelling has largely focused on the first two types of macrostructure, with little exploration of theme. In this paper I present a “thematic linking” model for authoring hypertext fiction, describe the implementation of this approach in the HypeDyn procedural hypertext fiction authoring tool (Mitchell, 2014), and discuss my initial experiences using this approach to authoring.

One way to apply the notion of “theme” to the authoring of hypertext fiction is to base the relationships between story elements, and the creation of links that the reader can follow, on thematic relationships. This is an extension of the concept of “sculptural hypertext”, where an author, rather than creating links, sets up rules that determine, at run-time, the connections between nodes (Bernstein et al., 2002). To explore this approach, I have created a modified ver- sion of HypeDyn that implements “thematic links” using a model for describing thematic relations within a text as proposed by Hargood (2011).

In Hargood’s model, stories consist of narrative atoms, or natoms, which are the smallest unit of a narrative. An example of a natom is a single posting on a social media site (photo, caption and tags). Natoms contain certain visible, computable elements or features, such as the captions or tags on a photo. These features denote certain motifs, which in turn connote themes within the story. A specific natom may be considered to have a numerical thematic quality with respect to a given theme. The thematic quality can be used to compare natoms and determine which are more closely related to the theme.

To create a story using thematic links, the author first creates a series of nodes. She then creates a thematic structure consisting of themes, motifs, and features. The features correspond to fragments of text within the nodes. Links are automatically generated from the current node to thematically similar nodes, as calculated using Hargood’s thematic model. Using this method, the author is essentially creating a set of indirect connections between nodes by means of the motifs and themes.

A number of challenges arise when using this approach. One issue is how to give the author some sense of how nodes will eventually be connected. Another challenge is how to allow for some form of thematic progression across the story – if links are always generated to thematically related nodes, the story may end up literally going in circles. Finally, there is the question of how the reader’s choices can be incorporated into the decisions the system makes as to how to generate the thematic links. The paper discusses these issues, grounded in my own experiences using thematic links to write a short, 50-node hypertext.