ENG 574: Are Good Texts Always Better? - Sarah Felicelli

Article Citation
McNamara, Danielle S., Eileen Kintsch, Nancy Butler Songer, and Walter Kintsch. 1996. “Are Good Texts Always Better? Interactions of Text Coherence, Background Knowledge, and Levels of Understanding in Learning from Text.” Cognition and Instruction 14 (1): 1-43.

Summary of Article
McNamara et al. conducted two studies to examine the relationships between text coherence, prior knowledge, and levels of understanding in reading comprehension of science text. Text coherence is related to semantics, or the meaning of the text of a document. In other words, the more a text makes sense to readers, the more coherent the text is considered to be.

In the first study (Study A), the researchers tested the comprehension of students who read an excerpt from a biology textbook. Three groups of students each read a different version of the same text: the original version (590 words), a revised version (821 words), and an expanded version (1,167 words). For the revised version, information was added to the text to clarify the subtopics; whereas for the expanded version, content was added to explain the connection between different concepts. All three texts were locally coherent (the content made sense at the level of the text); however, the revised and expanded texts were also globally coherent (the content made sense in relation to information beyond the text). Each student took a pre-test and a post-test. Both the pre- and post-tests had multiple choice, true-false, fill-in-the-blank, and short-answer questions as well as a sorting task that the students completed. For the sorting task, students arranged terminology cards into three categories (9-10).

Researchers found that students who read the revised version scored better on the post-test questions than those who read the original version. Students also improved on the sorting task. For the expanded text, in some cases, comprehension was not as high as for that of the revised text. One explanation suggested by the investigators was that the increased word count of the expanded text raised the level of difficulty for students with a reduced reading skill.

In the second study, the researchers also examined the influence of prior knowledge on the ability of students to recall information from a less coherent text. To test this, they created four versions of a text with the same content (heart disease) but with different levels of local and global coherence. For the high-coherent versions, they revised the text by:

• Replacing pronouns with noun phrases to decrease ambiguity.
• Adding descriptions to connect unfamiliar concepts with familiar ones
• Adding sentence connectives to clarify relationships
• “Replacing words to increase argument overlap (e.g., replacing person and cases with baby or babies).” (21).

Pre- and post-test questions differed from Study A in that they included text-based questions (based on one sentence in the text), elaborative-inference questions (based on content from the text and outside knowledge), bridging-inference questions (based on topics linked between two or more sentences), and problem-solving questions (based on information from the text that was described in a new context). A sorting task was included for Study B as well (22).

McNamara et al. found that for bridging-inference questions, problem-solving questions, and the sorting task, low-knowledge readers needed a high-coherent text to do well, whereas high-knowledge readers fared better with a low-coherent text. The investigators reasoned that for high-knowledge readers, a low-coherent text increased their comprehension, because they filled in the coherency gaps on their own and were more actively involved in learning the material (34).

Analysis:
Dr. Danielle S. McNamara is a professor with the Department of Psychology at University of Memphis. Her expertise is in cognitive psychology and her research interests include the processes of reading comprehension, memory, and knowledge acquisition. Drs. Eileen and Walter Kintsch are researchers with the Institute of Cognitive Science at the University of Colorado, Boulder, and Dr. Nancy Butler Songer is a professor at the University of Michigan who specializes in science education.

The literature review for this article is both thorough and beneficial to further understanding my research topic and the hypotheses for the studies. Both studies had participants who were between 6th and 10th grades. The first study had a sample size of 36 and the second had 56 participants. For both studies, the data analysis that was performed was appropriate for the study design (separate ANOVAs for a 3X2X2 design for Study A and separate ANOVAs for a 3X3 design for Study B).

Reflection
This article is appropriate for my research project as it directly relates to comprehension of science text. However, the research is somewhat dated—it was published in 1996. For my final project, I am considering the possibility of replicating a part of one of the studies done by the researchers in this article. The experimental designs for both studies are somewhat complex, so I may not be able to manage a replication in the short time period allotted for the ENG 574 final project. If I were to replicate one of these studies, I would change the population to college students to see if there is a difference in the results. Also, I would likely choose only one aspect of one of the two studies to simplify the study.