Stance, Navigation, & Reader Response in Expository Hypertext - Page 1 of 18
Stance, Navigation, & Reader Response in Expository Hypertext
Digital Literacies Research Group
Department of Reading and Language Arts, OaklandUniversity
Kris Allen, Lizabeth Guzniczak, Ledong Li, John McEneaney
{kjallen, laguznic, 11li, mceneane}@oakland.edu
ABSTRACT
This article reports on two studies investigating reader stance, navigation, and response in expository hypertext from a transactional perspective (Rosenblatt, 1994a, 1994b, 1995). The studies apply quantitative and graphic methods in exploring the influence of hypertext structure and reader stance on navigation and reader response. Subjects in the two studies include 69 and 147 mature readers, respectively, who were prompted to adopt either an efferent or an aesthetic stance reading a 36-node expository hypertext. Reading was followed by recall and essay writing tasks. Both studies adopted between-subjects designs with a nominal stance factor (aesthetic and efferent). Multivariate analyses were used to assess the influence of stance on sets of navigational and response variables, with follow-up analyses assessing the influence of stance on individual variables. Main effects for stance on both navigation and the essay response measures were noted, including a finding linking higher levels of understanding to readers adopting an aesthetic stance. Correlational analyses failed to discern associations that had previously been noted between navigation and reading outcome measures suggesting that interpretation of navigational measures must account for hypertext structures. Graphic analyses revealed wide variability in the navigational strategies of readers both across and within groups with distinctive patterns that related to reader strategy and stance. Exploratory analyses examining data collapsed across both studies reinforces prior findings and suggests node size may play a role in navigation and response as well.
Transactional theory (Rosenblatt, 1978, 1994, 1995) represents one of the most significant frameworks for conceptualizing reading and writing developed in the twentieth century and continues to have a powerful influence on literacy researchers and practitioners. Two fundamental concepts are at the core of transactional theory: 1) in contrast with other models that "locate" meaning in the text and conceive of reading as extracting that meaning, transactional theory defines meaning as an event or process that requires simultaneous attention to both reader and text, and 2) the understanding a reader creates depends in a critical way on the stance adopted, where the term stance refers to the purpose or general orientation of the reader. Stance is typically characterized as a continuum anchored at one side by an aesthetic orientation and at the other by an efferent orientation, where an efferent stance involves a reading goal tied to subsequent action and an aesthetic orientation focuses on a lived through experience (Rosenblatt, 1995, p. xvii) of the reading act itself.
Although researchers routinely acknowledge stance as a continuum and emphasize that any given reading event incorporates elements from both ends of this spectrum, both theoretical and empirical work has focused almost exclusively on the aesthetic stance, largely ignoring problems and issues specific to efferent reading. Even in those instances where there is an effort to study the efferent end of the continuum, literary narratives remain the dominant text type used in these studies, material associated with aesthetic rather than efferent objectives (Guzniczak, 2004). While a special focus on aesthetic reading is not, in itself, a problem it seems clear that research that adopts a one-sided view of the stance continuum may skew our efforts to define a truly general framework for thinking about the reading transaction (1978, pp. 22-47). Moreover, in more recent work, the claim that transactional theory is intended to support a general framework supporting both efferent and aesthetic orientations has become more explicit (Rosenblatt, 1994, p. 1069; Karolides, 1999), but even in these more recent elaborations the exploration of the efferent stance and its consequences fail to go beyond the limited analyses carried out in earlier work. Consideration of the aesthetic transaction ranges from the escapism of "trashy" fiction (1978, p. 143), to sophisticated critical interpretation of the Iliad (1978, p. 122), in addition to numerous other examples from American and British literature. In contrast with this, the only example of an explicitly efferent transaction, apart from references to generic types of reading material (e.g., "a newspaper, scientific text, or cookbook," 1978, p. 22), concerns reading a label to determine appropriate treatment for ingested poison (1978, pp. 23-24; 1994, p. 1066). The absence of more serious exploration of efferent transactions is, moreover, somewhat surprising given that important foundational work cited by Rosenblatt (i.e., Dewey, 1938 and Dewey & Bentley, 1949) focuses primarily on the efferent end of the transactional continuum. While Rosenblatt certainly is not misleading about her focus on aesthetic reading and provides a rationale for the commitments she makes, it seems reasonable to ask whether a general theory of reading can rely so heavily on one stance without risk of distortion or omission.
A second aspect of transactional theory (and reading theory in general) we seek to address concerns the disconnect between theories of literacy and theories developed to support many of the new literacy technologies that are changing the ways people read and write. Although transactional theory has traditionally been framed as a pragmatic natural-language phenomenology of reading, there are practical reasons for considering an approach that better aligns with technical frameworks used to define and describe online literacy environments from a computing perspective. One reason is the remarkable transformation of literacy practices since the early 1990s, with the widespread use of the Internet and computer-based literacy tools. These technologies challenge some of our most fundamental assumptions about readers and text (Landow, 1992; Reinking, 1998) and all are, to a greater or lesser extent, grounded in formal frameworks. Literacy theory, however, generally operates as an independent theoretical layer, floating "above" the theoretical substructures that support the new technologies of literacy that are having such a profound impact on literacy practice. While it would be both unwise and unproductive to demand that all of literacy theory be defined within a technical framework that also accounts for literacy technologies, it seems quite clear that there are important points of contact where literacy theory and underlying formal frameworks established by the new technologies of reading and writing can and should be explored and that this kind of study will help redefine literacy theory in ways that more adequately acknowledge the power and influence of technology in shaping literacy practice and theory.
Our studies address questions related to the issues of navigation, stance, and reader response in expository hypertext. This work focuses on the two aspects of transactional theory we have briefly reviewed above by adopting expository reading material and framing our work within a context that incorporates perspectives from both literacy studies and the technical domains that have produced hypertext technologies. A central concept supporting our work is the view that hypertext structure and a reader's navigational choices during a reading episode are expressions of the reading transaction and digital artifacts produced by these online technologies provide an important new research window on the reading transaction (McEneaney, 2001, 2003). This study explores graphical approaches to visualizing the reading transaction as a "path." We analyze reader paths in hypertext in several different ways. One form of analysis is based on a graphical method intended to illustrate both individual and group paths in a way that makes navigational patterns visually distinct. A second approach involves analyzing reader paths according to the content of reading material that appears to be driving navigational decision making. The third form of analysis relies on quantitative measures that are related to the visually distinct categories identified by graphic analysis. These path-specific structural measures have been described in detail elsewhere (McEneaney, 2003) and have been shown to correlate significantly with reading
outcome measures (McEneaney 2001). Our research is guided by three main research questions.
1)What is the influence of reader stance on hypertext navigation and response?
2)Are there associations between navigational patterns and reader response measures?
3)Can temporal structures be discerned in the online transaction that influence reader response?
Hypertext structure and navigation
Our analyses of navigation in hypertext distinguish three distinct types of structure that must be acknowledged in understanding the online transaction: virtual structures that specify what is possible, episodic structures, that specify outcomes of individual reading transactions, and emergent structures that specify broader shared structures that emerge from the accumulated transactions of multiple readers (McEneaney, 2003). Virtual structure is defined by the nodes and links that make up a hypertext and can be viewed as a property of the text itself. Episodic structure, on the other hand results from a specific transaction involving a reader and text, and corresponds to the structure a reader creates during reading. Finally, emergent structures are abstracted structure patterns, based on both episodic structures and agreed upon (i.e., warranted) methods for collecting, organizing, and representing those collective structures.
ToFrom / 0 / 1 / 2 / 3 / 4 / 5 / 6 / 7
0 / 0 / 0 / 0 / 1 / 0 / 1 / 0 / 0
1 / 0 / 0 / 0 / 0 / 0 / 0 / 1 / 0
2 / 0 / 0 / 0 / 0 / 1 / 0 / 0 / 0
3 / 1 / 0 / 0 / 0 / 1 / 0 / 0 / 0
4 / 0 / 0 / 0 / 0 / 0 / 0 / 0 / 1
5 / 0 / 0 / 1 / 0 / 0 / 0 / 1 / 0
6 / 0 / 0 / 0 / 0 / 0 / 0 / 0 / 1
7 / 0 / 1 / 1 / 1 / 0 / 0 / 0 / 0
A. Adjacency matrix for the digraph in B. /
B. A node and link representation of the adjacency matrix presented in A.
Figure 1. An adjacency matrix (A) and its corresponding graphical representation (B).
The framework we apply in assessing and visualizing reader navigation is based on the traditional node-and-link model of hypertext. Although not without drawbacks (Stotts & Furuta, 1989; Turine, de Oliveira, & Masiero, 1997), this model of hypertext has proven to be a useful conceptual framework from both theoretical and empirical perspectives. Central to this framework is the idea that a hypertext can be conceptualized as nodes that represent content and links that represent structure. At least part of the popularity of the node and link model can be attributed to two simple but powerful formalisms that support analysis in this model: adjacency matrices that are well suited to mathematical analysis, and labeled directed graphs (digraphs) that present structural information in a readily interpreted visual format. Briefly, an adjacency matrix is a table that records each link in a hypertext document. Typically, an adjacency matrix consists of a table of zeroes and ones with labeled rows and columns. A "1" in cell (a,b) indicates a direct link from node a to node b. A "0" in a cell indicates that there is no direct link between the two nodes. In Figure lA, for instance, the two entries of "l' in the first row indicate that there is a direct link from node 0 to nodes 3 and 5. Zeroes appear in all other positions in this row because no other direct links are present. Figure lB illustrates this hypertext, demonstrating that the document consists of eight nodes (numbered 0 - 7) and 13 links (note that the arrow connecting node 0 and node 3 is double-headed and thus represents two links, since it operates in both directions.)
What makes these formalisms powerful is that they retain mathematical and visual simplicity yet succeed in capturing important structural features that can be transformed to create other useful measures, including measures that can be used to assess and display the paths individual readers and groups of readers take when reading hypertext. Moreover, graphs and digraphs are widely used in hypertext research as a basis for visual analyses that can reveal structure difficult to discern in numerical formats (Chen, 1997; Mukherjea & Foley, 1995; McDonald, Paap, & McDonald, 1990). Two measures of special interest in previous related work are path compactness and path stratum (McEneaney, 2001). The purpose of these measures is to yield global assessments of navigational structure. Path compactness refers to the connectedness or complexity of a path structure, yielding values close to zero for simple, sparsely connected path structures and values close to one for densely connected path structures. Path stratum, on the other hand refers to the degree of linearity of a path, as indicated by the extent to which the path structure is organized so that certain nodes are always read before others. More linear paths have stratum values closer to one, while those less linear are closer to zero. The real significance of these measures, however, is that they appear to be related to other more traditional reading outcome measures such as recall and problem solving with text.
Recent work has, for example, explored the relationship between navigational choices readers make reading hypertext documents (i.e. web pages) and their success in applying their comprehension to problem-solving tasks (McEneaney, 2001, 2003). Results of this work indicate that navigational patterns and associated numerical measures can be interpreted as reflecting both user strategies in deciding what links to follow and relative success in cognitively modeling the domain represented by a hypertext. Subjects who adopted shallow, hierarchical search strategies that more accurately modeled the organization of the hypertext materials were more successful in their reading tasks than were those who adopted more linear paths (See Figure 2) and these navigational patterns were also evident when summary data of more and less-successful reader groups were analyzed. Moreover, the informal visual interpretation of these striking navigational patterns was quantitatively corroborated when path stratum and path compactness were found to correlate significantly with the comprehension measure used in the study (McEneaney, 2001). These measures were, however, originally designed for use with hierarchically structured hypertext. One goal of our second research question is to assess the extent to which these measures can be applied in hypertext with a non-hierarchical design structure.
Method
Previous work (e.g., Many, 1994; Hartmann, 1994; Rosenblatt, 1978, 1994) is unequivocal in supporting the claim that readers' perceptions about goals and purpose of reading have a significant impact both on reading process and outcomes. The intrusive nature of "think aloud" methods, however, in which readers are required to interrupt or recall the reading process to report on their decision making makes it difficult to generalize from such studies (Laing & Kahmi, 2002; Magliano, Trabasso, & Graesser, 1999). In contrast, in electronic reading environments real-time data about reader decision making and allocation of attention can be assessed in very direct and non-intrusive ways as a result of the interactive nature of link selection in hypertext. The work we describe capitalizes on the interactive potential of online reading environments, capturing this real-time data and applying it in addressing questions about the nature of reader stance and response.
Subjects in both studies were mature readers attending a medium-sized public mid-western university in the US. All subjects were education students who volunteered to participate in the study. Some students were provided extra credit from instructors for participating. Protocols were reviewed and approved by a university human subjects review board. Reading sessions began with subjects logging into a web server that provided access to experimental materials. This web server also collected information about the pages subjects chose to read and the amount of time (in milliseconds) each subject spent reading each page. Prior to reading, each subject was provided one of two different prompts. One prompt was more aesthetic in nature, directing the reader to focus on his or her personal response to the reading. The second of the two prompts was more efferent, directing readers to focus on the content and educational significance of the ideas presented. Subjects were allowed to take handwritten notes during the reading session. Following the reading session, subjects put away any notes they had made and completed a true-false recall measure. Following the recall measure, students were allowed to take out their notes and use them to write an essay responding to the reading and the prompt they were provided before the start of reading. At the conclusion of the experimental session, data collected during the reading session was written to a server-based file.
Recall scores were based on two forms of a true/false recall measure. Study 1 employed an assessment based on 72 items that required handwritten responses. Study 2, on the other hand, adopted a shorter 36-item assessment that was computer-based. All notes were handwritten but essays, like the recall measure, included both handwritten (in Study 1) and computer-based (in Study 2) responses that relied on keyboard input. Word counts were carried out for both essays and reader notes. Essays were also scored using two rubric-based systems that have been used in prior research (Cox & Many, cite). One scale provided a Stance Rating (SR) Score indicating the extent to which a response was more aesthetic or more efferent in nature. This measure was based on a 5-point scale with a "1" representing the most efferent response, a "5" representing the most aesthetic response and a "3" indicating a balanced reader response. The second essay scale was a level of understanding rating (UR). This measure was based on a three-point scale with a "1" assigned to essays limited to restatement of facts, a "2" for those essays that mixed facts with text-to-self, text-to-text, or text-to-world connections (CITE), and a "3' assigned to those essays that attained generalized understandings based on article content. Both measures were based on similar rubric-based measures employed by Cox and Many (1997, ...) in a series of studies exploring reader response in literary text. All essays were scored twice on each scale by two different raters with average scores used in subsequent analyses, except those documenting inter-rater reliability. Rubrics used in scoring essays are provided in the Appendix.
The hypertext used in the studies is an adaptation of an article published in Reading Online (McEneaney, 2000). This material was selected both because it provided an authentic expository hypertext originally written in hypertext format that had appeared in an online journal and because it provided a contrasting virtual structure to the hierarchically organized hypertexts adopted in previous studies (McEneaney, 1999, 2000, 2001). The adaptation used in Study 1 presented the article with only minor modifications to the introductory material orienting the reader to the hypertext format. The adaptation in Study 2 presented a shorter version of the article revised so that scrolling was not required to view the content of individual nodes. The article presents a brief history of hypertext in a 36-node hypertext organized in a grid format with columns in the grid structure representing historical eras and rows in the structure representing six different themes that are labeled to the right of the grid image (see Figure 3). The reader's location within the hypertext is indicated by an "x", providing an immediate visual cue for the reader's position within the hypertext structure. Headers at the top of each node indicate the era and theme of the node and navigational buttons at the upper right allow readers to navigate by historical era, topical theme, or along a predetermined path across all 36 nodes. Navigational data were temporarily stored in a browser cookie during reading and uploaded to a database at the conclusion of the reading session when readers clicked the "End Session" button. For students in Study 2 using the computer assessment format, clicking the "End Session" button also presented the online recall assessment and a text field for entering the response essay from the keyboard.