Lately we have been bombarded with prophecies about how hypertext is changing the literary scene irrevocably. "The book is dead," futurists proclaim, "long live the docuverse!" Hypertext, they say, makes the postmodern textual ideal a reality, empowers readers, and mimics the associative thought processes of the mind. Amid the discourse on reading literary hypertext, however, there are few examinations of the behaviour of actual readers.
A notable and interesting exception to this reality is Jane Yellowlees Douglas's discussion of how fourteen students in her writing class at New York University responded to Stuart Moulthrop's Storyspace version of Borges's "The Garden of Forking Paths." Douglas describes the bafflement expressed by her students when presented with this text:
Confused by a multiplicity of narrative strands in which they could encounter a character dead in one place and very much alive and ambulatory in the next, the readers of "Forking Paths" lacked any tangible sense of a macrostructure which could give significance to the elements they encountered in each individual narrative segment. (par. 33)
The experience of Douglas's students suggests that readers of literary hypertext, far from being empowered, are at times placed in a Faustian predicament in that their choices are often more limited than they appear. Links are not programmed at random, after all; rather, their presence and their direction are as integral to the artistry of the work as is content. How might we describe the activity of reading in such a controlled textual environment? The words of Mephistopheles to Faustus come to mind: "When thou tookest the book/ To view the scriptures, then I turned the leaves/ And led thine eye" (Doctor Faustus V.ii.89-91).
By way of approaching an understanding of how readers experience hypertext, and what reading strategies they might employ, this paper describes a two-part study of seventy hypertext readers. Participants read an electronic adaptation of a modernist short story, Elizabeth Bowen's "The Demon Lover." The story was presented to them in one of two electronic forms, both of which were constructed using the HyperWriter! authoring system. Subjects in the control group read the story in a structurally linear format, activating a "next" link to move from paragraph to paragraph. The hypertext group read the same text in what I call simulated self-navigating format. As in the linear text, the story was broken down into its paragraph divisions and presented as a series of separate nodes; having read one lexia, however, readers were required to choose between two or three embedded links in order to progress to another. A simulation of multi-sequential text was thereby effected. The "Output Report Action" of HyperWriter! enabled the automated collection of reading times and link choices. Following reading, each participant self-recorded their observations about the experience of reading the text and completed both a demographic survey and a questionnaire about their reading preferences.
The first phase of the study involved determining patterns in the transcript data. To this end, a taxonomy of reader experience was developed and the transcripts were tagged to enable searches for frequency and co-occurrence of particular concepts. Searches of the transcripts in WordSmith Tools revealed a high level of disorientation amongst simulation readers, many of whom expressed frustration with the unresolved nature of the ending and were puzzled or frustrated by links. Further, several simulation readers reported a feeling of agitation accompanied by a desire progressively to increase their speed of reading.
To confirm this preliminary indication that the hypertext readers modified their reading in some substantial ways, actual reading speeds and link choices were examined in a second phase of the study. The findings revealed marked differences in reading strategies between the groups. For example, the hypertext readers took nearly eight seconds longer on average per section than the linear readers (linear: M = 35.43 secs; hypertext: M = 43.22), a significant difference, t(23) = 3.823 p < .01. Reader pace, however, changed dramatically in the course of reading: linear readers read progressively faster in the first half of the story (r(9) = -.470, p < .05) but not the second half; this is reversed for hypertext readers, who read noticeably faster towards the end of the story (r(8) = .992, p < .01). Their progressively increased reading speeds in combination with evidence from the transcripts suggests that the attention of simulation readers was diverted to the surface features of the text, and that their reading patterns became increasingly fragmented as they advanced through the story. Other analyses will be reported in the paper.
Although it is impossible to project with assurance from this preliminary study, these findings suggest that hypertext readers do not usually experience feelings of empowerment, control and freedom, nor do they think of themselves as "co-authors" or "collaborators." Instead, they struggle at times to become oriented and are often confused when presented with links. Furthermore, in some instances readers demonstrate agitation and an inability to engage with the narrative, resulting in fragmented understanding. Clearly there is a wide gap between much theoretical discourse and the reality of reader experience as evidenced in this study.