Connie Gray PMCH 6481: Assignment 3 Nov. 12, 2004 A Literature Review of the Effect of Text Format on Readers' Reading Comprehension Topic Sentence: What is the effect of text format (traditional linear text vs. non-linear hypertext) on reading comprehension? Introduction Emerging technologies are being used more and more in our classrooms. With the advent of hypermedia and the internet, students are now exposed to written information in a new way. A proliferation of texts in hypertext format are now being produced. Traditional print, what we see in traditional textbooks and learning materials, presents information in a linear manner. The information is offered to the readers in an order decreed by the writer. Readers typically progress through their reading of traditional print from beginning to end in sequential order. In contrast, hypertext is nonlinear and provides more flexibility to the reader in choosing where to go next. The reader experiences more choices as they navigate through the material, clicking on hyperlinks (links to other pages of information) based on their needs and interests. Research comparing linear and hypertext has produced varied results. Studies have explored avenues such as the impact of readers' prior knowledge, their cognitive abilities, working memory capacity, and the makeup of the text itself on reading comprehension. Research Question: This literature review revolves around the following research question: "What is the effect of text format (traditional linear text vs. non-linear hypertext), if any, on reading?" Idealized Independent Variable: The idealized active independent variable explored in this literature review is text format, operationally defined as the categorical variables: traditional text (linear) or hypertext (nonlinear, structured or networked hypertext). Idealized Dependent Variable: The idealized dependent variable explored is reading comprehension. The dependent variable is ambiguous by its very nature and therefore not easily defined in operational terms. Much research has been done on linear text reading comprehension; user models have been used to determine what information will be learned from linear text. For the scope of this literature review, however, participants' reading comprehension will be limited to and evidenced by their scores on comprehension assessments. Idealized Directional Hypothesis: Participants using non-linear structured hypertext will perform better on written reading comprehension assessments than those using non-linear networked hypertext or those using traditional linear text. 1 Connie Gray PMCH 6481: Assignment 3 Nov. 12, 2004 A Literature Review of the Effect of Text Format on Readers' Reading Comprehension Discussion of Articles Each of the studies reviewed examined more than one hypothesis in multiple experimental setups. For the purpose of this review, I will evaluate and analyze only the data and results which correspond to the idealized independent and dependent variables. Differences in how the IVs were operationally defined: All the research studies consistently used traditional-style text format as their linear category of the independent variable. However, only the Lee and Tedder (2003) and Calisir and Gurel (2003) studies used hypertext categories similar to that of the idealized independent variable. Lee and Tedder classified their hypertext format as 'structured' or 'networked' while Calisir and Gurel grouped their hypertext format as 'hierarchical' or 'mixed'. The studies by Lowry (2004) and McEneaney (2003), are only semi-comparable to the idealized independent variable. Both of these studies used only 'non-linear' to define their version of hypertext. No mention was made as to the topology of the hypertext itself. The other two studies (Wenger, 1996) and (Kasper, 2003) were broader and more specific in their categorization of hypertext, using up to 4 and 5 categories of hypertext respectively. Differences in how the DVs were measured: All of the studies measured reading comprehension by analyzing scores on comprehension tests. Due to the nature of the experiments, i.e. tailoring their questions to the specific content of their linear and hypertexts, none of the question sets were identical. The Wenger study (1996), collected both online performance measures (reading rate, concurrent task accuracy, and response time and time required to make sequencing decisions) and reading outcome measures (test for recall, test for comprehension, and test for memory retention of text structure). However, the McEneaney study (2003), used two different question sets to measure reading comprehension instead of one. Another significant difference in the way the dependent variables were measured is within the question sets themselves. Most of the studies used short multiple choice recall tests, one study used both multiple choice and fill-in-the blank items, (Calisir, F. and Gurel, Z., 2003), and one study used short answer question sets (McEneaney, J., 2003). Sources of internal validity compromise (e.g., problems with the design and execution of the study): All of the studies contain questionable aspects of internal validity. Of primary concern in all the studies is the reliability of the recall tests. The methodology used to create the questions sets were remains vague. From the sample question sets provided by the researchers, it is apparent the majority of the questions used to assess reading comprehension could all be considered at the lowest level of bloom's taxonomy, knowledge and recall, and very few questions were at the comprehension level. The 2 Connie Gray PMCH 6481: Assignment 3 Nov. 12, 2004 A Literature Review of the Effect of Text Format on Readers' Reading Comprehension 'purpose' for reading may also have had an effect on the results. The participants were 'rewarded' with course credit in order to promote their motivation to do their best. However, reading for pleasure and reading for finding information are very distinct and may influence memory recall. In addition, the only study in which the participants were told to study the documents with the goal of comprehension in mind was the Calisir (2003) study. Other types of tasks or reading purposes as perceived by the participant may have had an effect on results. Another common threat to internal validity among all of the studies (of which only the McEneaney, (2003) study specifically discussed) may have been the participants' familiarity with technology tools and practices. The researches failed to mention if they trained their participants in navigating through hypertext. A third common concern with respect to internal validity is the amount of time readers were allowed to spend reading the content. Slower readers and unmotivated readers' scores may be negatively affected if they are not required to read the entirety of the text. Only the Wenger (1996) study discussed whether subjects read all of the nodes (ie links' or hyperlinks' content) within the hypertext and even then, it was mentioned that their subjects covered on average only 85% of the available nodes. Furthermore, Lowry, W. (2004) and McEneaney, J. (2003) failed to identify what format of hypertext they used in their respective studies. Other types of topologies (e.g. structured, networked, or mixed hypertext) may have had different effects on results. A final major danger to internal validity was found in the Kasper (2003) study. While there were different categories of hypertext in their study, the researcher designed each type of hypertext to be focused on a different subject matter instead of designing different types of hypertexts focused on the same subject. Some subjects may have been more difficult for students to comprehend and may have influenced their performance and reactions to the different types of hypertext used Differences in samples, populations, and assessments of external validity: Limitations which could affect the external validity of the results and which should be taken into account for future studies include the limited size of the sample pool. Here again, all five studies shared a common thread -- all of the studies were limited by their populations and sample size; the Lee and Tedder (1993), study severely so. After dividing their sample size into three conditions (traditional-linear, structured hypertext, and networked hypertext) and then further subdividing those participants into low, medium, and high working memory span groups, certain categories had too few a sample size. Similarly in the McEneary, (2003) study, the small sample size resulted in a 3 Connie Gray PMCH 6481: Assignment 3 Nov. 12, 2004 A Literature Review of the Effect of Text Format on Readers' Reading Comprehension statistical power below the recommended level). For the most part, the studies were carried out with college level students as participants. A more representative sample of multiple ability reading levels in a more diverse population needs to be done in order to increase external validity. According to Lowry (2004), younger audiences may focus more on format and presentation of the material than on the actual content of the text. Implications of the degree to which the studies support or deny the hypothesis: Analysis of the studies provided contradictory results. Results of the Calisir and Gurel (2003) and the Lowry (2004) concluded that readers of the non-linear story were no more successful than were readers of the linear story at recalling story information. Calisir and Gurel further concluded that there presented no significant difference between knowledgeable and non-knowledgeable participants in the hierarchical and the mixed conditions. Conversely, Lee and Tedder (2003), and McEneany (2003), found that higher recall scores were achieved by participants reading traditional linear text. Alternatively, the studies conducted by Kasper (2003), and Wenger (1996) both supported the idealized hypothesis. Kasper's study indicated that all forms of hypertext produced higher scores on comprehension tests than did print texts. Wenger's study also agreed that comprehension was higher for participants using hypertexts than linear texts but with the added caveat that the comprehension advantage for hypertext was apparent only for the description texts presented as hypertexts, with the description texts presented as linear text actually showing lower performance than causation texts presented as hypertext. Within the topology of different types of hypertexts, results were contradictory as well. Lee and Tedder concluded no significant different between structured and networked hypertext when total reading time was taken into consideration, while Kasper concluded that student feedback and performance data indicated that controlled hypertexts with glosses led to the highest level of text comprehension. Final Conclusions The studies all investigate the effect that hypertext has on the reader's comprehension of the material, yet the results significantly with each other. Assessing reading comprehension is difficult matter. Many years of research have been spent on the evaluation of readers' comprehension, all of it dealing with linear traditional text. The contradictory results found in the studies reviewed with respect to reading comprehension as it pertains to nonlinear hypertext only serves to prove how much more research in this field is needed. Research needs to be done in order to determine where and in what scenarios hypertext will be more successful than linear text. Some features of hypertext which will enable it to succeed over traditional print text include: the ability for readers to find relevant information in an efficient manner and search through the entire body of the text with ease, the ability to represent textual information that is not as easily presented in linear form, and its ability to provide additional information to the reader so that the text 4 Connie Gray PMCH 6481: Assignment 3 Nov. 12, 2004 A Literature Review of the Effect of Text Format on Readers' Reading Comprehension automatically contains information that is more appropriate to the reader's level of reading (Foltz, 1996). References: Calisir, F. and Gurel, Z. (2003). Influence of Text Structure and Prior Knowledge of the Learner on Reading Comprehension, Browsing, and Perceived Control. Computers in Human Behavior, 19(2), 135-145. Foltz, P.W. (1996) Comprehension, Coherence and Strategies in Hpyertext and Linear text. In Rouet, J.-F., Levonen, J.J., Dillon, A.P. & Spiro, R.J. (Eds.) Hypertext and Cognition. Hillsdale, NJ: Lawrence Erlbaum Associates. Kasper, L. (2003). Interactive Hypertext and the Development of ESL students' Reading Skills. The Reading Matrix, 3(3). Lee, MJ and Tedder, MC. (2003). The Effects of Three Different Computer Texts on Readers' Recall: Based on Working Memory Capacity. Computers in Human Behavior, 19(6), 767-783. Lowry, W. (2004). More Control, But Not Clarity in Non-linear Web Stories. Newspaper Research Journal. 25(2), p83. McEneaney, J. (2003). Does Hypertext Disadvantage Less Able Readers? Journal of Educational and Computing Research, 29(1), 1-12. Wenger, M. (1996). Comprehension and Retention of Non-linear Texts: Considerations of Working Memory and Material-Appropriate Processing. The American Journal of Psychology, 109(1), 93-131. 5 Connie Gray PMCH 6481: Assignment 3 Nov. 12, 2004 A Literature Review of the Effect of Text Format on Readers' Reading Comprehension Author(s) Lowry, W. (2004). More Control, But Not Clarity in Non-linear Web Stories. Newspaper Research Journal. 25(2), p83. RQ: What is the relationship, if any between the linearity of format of the Web story and the degree of perceived credibility by readers? Kasper, L. (2003). Interactive Hypertext and the Development of ESL students' Reading Skills. The Reading Matrix, 3(3). RQ: Study design to explore and asses the possible effects of hypertext on the development of ESL students' reading skills. Independent Variable The independent variable is a categorical variable with the two nominal categories: Level 1: Linear structured Web story Level 2: Non-linear structured Web story The independent variable consists of five hypertext conditions: Level 1: glosses (links provide pop up vocabulary) Level 2: controlled hypertexts (links lead to predetermined and limited number of texts on the topic Level 3: free hypertexts (students read main text and are directed to freely explore the internet for other texts related to the topic Level 4: controlled hypertexts with glosses Level 5: free hypertexts with glosses Dependent Variables Hypothesis Tested Threats to Validity Hypothesis Supported? Subjects' knowledge acquisition or ability to recall content from the story – measured by scores on question set Readers of nonlinear Web stories will demonstrate significantly greater knowledge acquisition than will readers of linear Web stories. The internal validity of the study may have been affected. The researchers controlled the amount of time readers were allowed to spend reading the story instead of allowing them as much time as they needed to process and read the information. This would negatively affect the results of slower readers. No. Readers of the nonlinear story were no more successful than were readers of the linear story at recalling story information. Student reading comprehension – measured by scores on comprehension tests What is the role of hypertext in developing ESL students' reading skills through a comparison of students' performance with different types of hypertexts. And 1 control: Printed Text Lee, MJ and Tedder, MC. (2003). The Effects of Three Different Computer Texts on Readers' Recall: Based on Working Memory Capacity. Computers in For Hypothesis 1, 2: The independent variable is a categorical variable with three nominal categories: Level 1: Participants Assess and evaluate the effects of different types of hypertext on reading comprehension and performance The dependent variable being investigated is an ordinal variable studying: The number of correct H1. The participants reading from hypertext documents will External validity of the study is compromised by the following factors: 1. Nature of the sample, which consisted entirely of undergraduate senior communications students. The researchers state that younger audiences may focus more on format and presentation of the material than on the actual content of the story. Internal validity is questionable in this study. The research designed each type of hypertext to be focused on a different subject matter instead of designing different types of hypertexts focused on the same subject. Some subjects may have been more difficult for students to comprehend and may have influenced their performance and reactions to the different types of hypertext used. Overall, the validity and reliability of the data is questionable. Of primary concern is the reliability of the recall test. No mention of how the test was created was mentioned. Yes. All forms of hypertext produced higher scores on comprehension tests than did print texts. Student feedback and performance data indicated that controlled hypertexts with glosses led to the highest level of text comprehension. Overall, this study indicates that the structure and presentation of text influence how well information is recalled. 6 Connie Gray PMCH 6481: Assignment 3 Nov. 12, 2004 A Literature Review of the Effect of Text Format on Readers' Reading Comprehension Author(s) Human Behavior, 19(6), 767-783. RQ: How does the presentation of the same information in different types of text formats, (the traditional linear text, the structured hypertext, and the networked hypertext,) affect a reader's recall based on his or her working memory capacity? Wenger, M. (1996). Comprehension and Retention of Non-linear Texts: Considerations of Working Memory and Material-Appropriate Processing. The American Journal of Psychology, 109(1), 93-131. RQ: What are the specific process that support comprehension and retention of hypertext information Independent Variable randomly assigned to traditional text, Level 2: Participants randomly assigned to structured hypertext, and Level 3: Participants randomly assigned to networked hypertext. The independent variable is a categorical variable with the following nominal categories: Level 1: Three linear format causation texts Level 2: Three hypertext format causation texts Level 3: Three linear format description texts Level 4: Three hypertext format Dependent Variables answers on the recall test – measured by a sixteen question recall multiple choice factual recall test Hypothesis Tested have lower recall scores than those reading identical information from traditional lineartext documents. H2. The participants reading from the structured hypertext will have higher recall scores than those reading identical information from the networked hypertext. Collected both online performance measures (reading rate, concurrent task accuracy, and response time and time required to make sequencing decisions and Reading outcome measures (recall, comprehension, and memory for text structure) Hypertext will impose more demands for relational processing than would linear text. Relational vs. item-specific processing Threats to Validity Hypothesis Supported? This study had several limitations which could affect the external validity of the results and which should be taken into account for future studies. 1. The sample size. After dividing the sample size into three conditions (traditionallinear, structured hypertext, and networked hypertext) and then further subdividing those participants into low, medium, and high working memory span groups, certain categories had too few a sample size. 2. The study only examined simple memory factual recall. Learning/memory is much more complex. The effects of hypertext may be different for higher cognitive thinking, i.e., synthesis or problem solving tasks. 3. The treatment of the information is not necessarily how one would find it in a real world setting. All hypertext links were limited and focused within the same document. Such is not true for real world Web pages. 4. The 'purpose' for reading may also have had an effect on the results. The students were 'rewarded' with credit in order to promote participant motivation to do their best. However, reading for pleasure and reading for finding information are very distinct and may influence memory recall. Internal validity is questionable in this study. No guarantee or evidence was provided that subjects read all of the nodes within the hypertext. It was only mentioned that the subjects covered on average 85% of the available nodes) Text format had a significant effect on recall scores. External validity may be compromised by the selection of participants. All subjects participated in the study in order to complete course requirements. It is not said if the Analysis of the data shows hypothesis 1 is supported. Higher recall scores were achieved by participants reading traditional linear text over those participants who read the structured hypertext. Analysis of the data shows hypothesis 2 is not supported. Higher recall scores were achieved by participants reading networked hypertext over those participants who read structured hypertext. However, this difference disappeared when total reading time was considered. Comprehension for hypertext was higher when reading with the spatial load rather than the digital load, whereas the opposite was true for linear texts Comprehension was higher for the description texts than for the 7 Connie Gray PMCH 6481: Assignment 3 Nov. 12, 2004 A Literature Review of the Effect of Text Format on Readers' Reading Comprehension Author(s) and the manner in which those processes different from those involved in the comprehension and retention of traditional linear text? Independent Variable Dependent Variables Hypothesis Tested description texts Threats to Validity selection was random or if students volunteered to be subjects. Test for recall Test for comprehension RQ: What is the relationship between print and hypertext reading skills in university students? Calisir, F. and Gurel, Z. (2003). Influence of Text Structure and The independent variable is a categorical variable with the two nominal categories: Level 1: Printed version of the handbook Level 2: Hypertext version of the handbook The independent variable is a categorical variable with the three Reading comprehension as measured by results to two sets of academic advising questions Reading comprehension (as measured by score by causation texts Results indicate that comprehension was higher for hypertexts than for linear texts. However, comprehension advantage for hypertext was apparent only for the description texts present as hypertexts, with the description texts presented as linear text actually showing lower performance than causation texts presented as hypertext Test for retention of text structure McEneaney, J. (2003). Does Hypertext Disadvantage Less Able Readers? Journal of Educational and Computing Research, 29(1), 1-12. Hypothesis Supported? Given the same reading material, do 'less able' readers make more effective use of hypertext or traditional print formats in a reading and questionanswering task? The internal validity of the study may have been affected by participants' familiarity with technology tools and practices. No mention of participants' level of comfort with using the technology was mentioned. The researches failed to 'train' their participants in their use. How does text structure and prior Only tested two types of hypertext topologies (hierarchical and mixed). Other types of 'Less able' readers found the hypertext version of the advising handbook more difficult to use than the print version. Limitations which could affect the external validity of the results and which should be taken into account for future studies include the limited size of the sample pool. (The small sample size resulted in a statistical power below the recommended level) and the question set used to assess reading comprehension (all were at the lowest level of bloom's taxonomy, knowledge, and very few questions were at the comprehension level) In linear text, knowledgeable 8 Connie Gray PMCH 6481: Assignment 3 Nov. 12, 2004 A Literature Review of the Effect of Text Format on Readers' Reading Comprehension Author(s) Prior Knowledge of the Learner on Reading Comprehension, Browsing, and Perceived Control. Computers in Human Behavior, 19(2), 135-145. RQ: What is the influence of text structure and prior knowledge of the learner on reading comprehension, browsing, and perceived control? Independent Variable nominal categories: Level 1: Tradition Linear Text Level 2: Hierarchical hypertext Level 3: Mixed Hypertext Dependent Variables Hypothesis Tested Threats to Validity 16 multiple choice and 22 fill-in-the-blank items based on content on all the nodes in the hypertext readings) knowledge of the learner affect reading comprehension, browsing, and perceived control? topologies (e.g. nonlinear grid and generative topologies) may result in different effects on reading comprehension, browsing, and perceived control. Number of different nodes opened during reading Perceived control as measured by questionnaire Participants were told to study the documents with the goal of comprehension in mind. Other types of tasks or reading purposes may have an effect on results. Population consisted of only graduate students at Istanbul Technical University. A more representative sample of multiple ability reading levels in a more diverse population needs to be done in order to increase external validity. Hypothesis Supported? participants ' performance was better than that of nonknowledgeable participants. There was no significant difference between knowledgeable and nonknowledgeable participants in the hierarchical and the mixed conditions. 9