Introduction to the Psychology of Science Text Comprehension
Arthur C. Graesser
University of Memphis
José A. León
Universidad Autónoma de Madrid
José Otero
Universidad de Alcalá
It is hardly a secret that students find most science texts very difficult to comprehend and that there are several reasons for these difficulties. The text is loaded with technical terms that need to be deciphered and memorized. There are complex mechanisms with multiple components, attributes of components, relations between components, and dynamic processes that flow throughout the system. Scientists frequently use a mathematical language, with symbols and formulas that are difficult to ground in everyday experience and that often require extreme precision. It is virtually impossible to form a mental image of some of the mechanisms without distorting the integrity of the system. Moreover, textbook authors often do not provide enough cues for readers to create coherent representations of information in science texts.
The problems are especially important for readers with poor scientific knowledge. In fact, all of the difficulties are exacerbated by the fact that most students have minimal background knowledge about science and therefore need to build an understanding nearly from scratch. Or, alternatively, they have incorrect knowledge that interferes with the scientific concepts and principles presented in textbooks. And of course, the complexity of scientific theories is increasing dramatically, year by year. As a consequence of all this, students frequently develop negative epistemic attitudes toward science texts and think of them as containing incomprehensible information. These attitudes negatively influence their text-processing strategies, in a continuing downward spiral.
All of these difficulties explain why reading science textbooks is difficult and why it has become difficult to entice students to major in science. The process of learning science is a challenge. Reading scientific text is a struggle that takes effort and concentration. Science texts are not a quick read.
School systems have periodically tried to meet the challenge by adopting radical pedagogical approaches. For example, the “physics first” approach reverses the order in which the different sciences are delivered in the school curriculum. The traditional order has been biology, then chemistry, then physics. The reason for this ordering allegedly is that biology has a high load on memory, but few exceptionally difficult conceptualizations that require a high IQ to master. So students keep busy memorizing parts of the anatomy and detailed taxonomies of animals and plants with exotic, lengthy Latin expressions. The utility of mastering precise genus and species labels is not exactly obvious and is rarely integrated with a deeper understanding of biology, but it does have a good side effect of promoting memorization and organization skills. Most of the fundamental mechanisms in biology are easier to grasp than those mechanisms in the sister sciences, so it makes sense to place biology earlier in the curriculum. In contrast, physics has the opposite profile: It is low on memorization and its key conceptualizations are difficult to master. Therefore, physics should come late. The problem with this curriculum plan is that students with a talent for science get turned off by all of the memorization in biology. A good scientific mind prefers to ask questions, generate hypotheses, play “what-if” games, experiment, test hypotheses, struggle with conflicting results, and become engaged in a host of other forms of reasoning and problem solving. Many scientific minds get turned off by a heavy dose of memorization, so unfortunately they never go into science. The physics-first approach tries to fix this problem by reversing the order of sciences in the curriculum: physics, then chemistry, then biology. So students quickly get started with a physics lab where they can experiment and build an inquiring scientific mind. The essence of the scientific mind is cultivated early and is not clouded by a horrendous exercise of memorization. The effectiveness of the physics-first approach is currently being evaluated, but some reports suggest that it significantly increases the number of science majors.
Another radical method of pedagogy has entirely discontinued science textbooks in the classroom and laboratories. The vision is to get the students to actively experiment in the laboratory, to build inquiring minds, and not to have them accept the textbook knowledge as gospel. This “delete the textbook” approach is perhaps more appealing when literacy levels are extremely low and the quality of textbooks is extremely poor. However, many researchers have been skeptical of the removal of the textbook from the science curriculum. There are times when students need to spend hours concentrating on textbook content until they master the difficult core concepts and mechanisms in a science, without getting distracted by the mundane practices of assembling equipment, collecting observations, and recording numbers in tables and charts. The key challenge is to arrange the learning environment so that the right text is available to the right student at the right time.
Nevertheless, the primary inspiration of this edited book does not really lie in the arena of science curriculum reform. Most of the authors in this book are researchers in cognitive science, discourse processing, and education who are building models of text comprehension. Our goal is to understand how children and adults construct meaning representations while they read and study texts. We develop theoretical models of the comprehension process and test the predictions of the model by collecting empirical data from readers. Some of the data tap the process of comprehension while text is read online (i.e., during reading). Examples of online measures include think-aloud protocols, sentence-reading times, the time to name test words aloud, and the timing and patterns of eye movements. Other data involve off-line measures that tap the result of comprehension, several minutes, hours, or days after comprehension is finished. Examples of off-line measures are recall tests, recognition tests on words or sentences, summaries of texts, question answering, and ratings of the importance of text constituents. A good theoretical model of comprehension can accurately account for rich patterns of data that include both online and off-line measures.
There are several reasons why science texts have attracted the attention of the comprehension researchers in this volume. One salient reason is that we can investigate comprehension under conditions in which comprehension is extremely difficult. As discussed earlier, scientific texts are difficult to understand at a deep level so these texts provide an interesting test case when the challenges of comprehension are pushed to the limit. Early research on comprehension focused on folktales, stories, everyday scripts, and other forms of narrative discourse that are easy to comprehend—the other end of the continuum on comprehension difficulty (Bruner, 1986; Graesser, Singer, & Trabasso, 1994; Mandler, 1984; Schank, 1999). Narrative is easy to comprehend because the content is very similar to the setting, actions, events, and social world we experience in everyday life. However, researchers in discourse comprehension have advocated moving from an emphasis on the study of narratives toward programmatic research on exposition (Lorch & van den Broek, 1997). That includes the development of theories of the structure and processing of science texts.
A second reason to study scientific texts is that there are more individual differences in comprehension processes among readers. Readers dramatically vary in their knowledge of the subject matter, their cognitive strategies of coping with exceptionally difficult content, their criteria in what it means to comprehend, and their motivation to persevere in mastering the science content. A good comprehension of scientific discourse fundamentally requires an excellent domain of highly specialized language, discourse, and world knowledge (Lemke, 1990; McKeown, Beck, Sinatra, & Loxterman, 1992; Means & Voss, 1985). In contrast, there is more uniformity among adult readers when they comprehend narrative text, at least narratives that do not have sophisticated literary forms (Graesser, Kassler, Kreuz, & McLain-Allen, 1998).
A third reason for investigating science texts is that the content of the material is useful for the readers to master. The content is not arbitrary or trivial, as in the case of much of the text materials that are written by experimental psychologists. Promoting science education fits a prominent mission in virtually all countries and cultures. Science textbooks have obviously played an important role in this endeavor. Yager (1983) reported that over 90% of all science teachers in the United States used a textbook 95% of the time. The importance of textbooks as a component of science instruction has also been advocated by other researchers (Chiapetta, Sethna, & Fillman, 1991; Gottfried & Kyle, 1992; Yore, 1991), in spite of the trend to minimize textbooks in some circles in science education.
A fourth reason for studying scientific text is because this genre of text has a distinctive way of organizing and explaining material. It is frequently assumed that coherence and comprehension are closely related. Under most, but not all circumstances, a coherently organized text facilitates the readers’ comprehension and subsequent task performance. However, sometimes the text per se is not sufficient for conveying the complex systems in mechanical, biological, or physical systems. The text needs to be enriched by adjunct illustrations, diagrams, tables, figures, photographs, and so on. Furthermore, in this electronic age, there are multimedia, hypermedia, simulation, and other computer technologies that allegedly facilitate more active learning and hopefully deeper comprehension. However, there is very little empirical research on the effectiveness of these nontextual technologies, so this is an important direction for future research.
WHAT IS SCIENCE TEXT?
We intentionally define science text very broadly in this volume. There is a broad definition of science and a broad definition of what falls under the umbrella of a scientific text genre. Regarding a definition of science, we adopt the natural category that is recognized in the National Science Foundation as SMET, which stands for science, mathematics, engineering, and technology. Our definition is compatible with Parker’s definition in the Concise Encyclopedia of Science and Technology (1994):
Science … is characterized by the possibility of making precise statements which are susceptible of some sort of check or proof. This often implies that the situations with which the special science is concerned can be made to recur in order to submit themselves to check, although this is by no means always the case. There are observational sciences such as astronomy or geology in which repetition of a situation at will is intrinsically impossible, and the possible precision is limited to precision of description. (p. 1661)
According to Parker, technology is a part of science, as described in the following:
Technology is a systematic knowledge and action, usually of industrial processes but applicable to any recurrent activity. Technology is closely related to science and to engineering. Science deals with humans’ understanding of the real world about them—the inherent properties of space, matter, energy, and their interactions. Engineering is the application of objective knowledge to the creation of plans, designs, and means for achieving desired objectives. Technology deals with the tools and techniques for carrying out the plans. (p. 1876)
The status of mathematics is perhaps on the edge of these definitions and is not directly addressed in this edited volume. However, all forms of science, engineering, and technology embrace some form of mathematics, which perhaps explains its inclusion in the SMET program of the National Science Foundation.
Our definition of the scientific text genre embraces several rhetorical forms and media. There are academic textbooks, scientific journal articles, technical manuals, magazine and newspaper reports tailored for the general public, information brochures for the public, and electronic multimedia on the Web and CD-ROM. The material is prepared by the author with the primary role of the diffusion of new knowledge about science. The chapter in this volume by Goldman and Bisanz presents a large landscape of science texts and their discourse functions. The chapter by Chambliss describes a theoretical framework for designing textbooks that integrate curriculum, instruction, and comprehensibility. Nearly all science texts are in the expository genre because they are written to explain and describe to the reader new content that has a foundation in truth and/or empirical evidence. However, some forms have a layer of persuasion, such as when a researcher is arguing with colleagues that a particular scientific claim is true or a particular scientific theory has merit. Scientific texts may also be in the narrative genre, as in the case of science history. It is widely acknowledged that many texts do not crisply fall into the traditional genre umbrellas of exposition, persuasion, narrative, and description (Brooks & Warren, 1972).
THE PRESENTATION AND PROCESSING OF SCIENTIFIC TEXT
The content of scientific texts has multiple levels of representation, but the most important split is between shallow and deep knowledge. Shallow knowledge consists of explicitly mentioned ideas in a text that refer to: lists of concepts, a handful of simple facts or properties of each concept, simple definitions of key terms, and major steps in a procedure (not the detailed steps). Deep knowledge consists of coherent explanations of the material that fortify the learner for generating inferences, solving problems, making decisions, integrating ideas, synthesizing new ideas, decomposing ideas into subparts, forecasting future occurrences in a system, and applying knowledge to practical situations. Deep knowledge is presumably needed to articulate and manipulate symbols, formal expressions, and quantities, although some individuals can master these skills after extensive practice without deep mastery. Deep knowledge is essential for handling challenges and obstacles because there is a need to understand how mechanisms work and to generate and implement novel plans. Explanations are central to deep knowledge, whether the explanations consist of logical justifications, causal networks, ...