The Tall Rob Report: van den Broek, P., Young, M., Tzeng, Y., & Linderholm, T. (1999). The landscape model of reading: Inferences and the online construction of a memory representation

Conflicts on the Internet stem from a mutual misunderstanding of a shared text. At some point, one of the parties to the discussion will suggest that one of the other parties lacks sufficient reading comprehension or perhaps was not paying attention during reading lessons in school. Here is a reheated annotated bibliography from an independent study section on reading comprehension that I completed during my Ph.D. in reading and literacy in early and middle childhood about ten years ago. I hope you find it useful in locating the possible reasons for your disagreements with others online and off.

Richard Anderson and P. David Pearson’s chapter in the first volume of the Handbook of Reading Research (1984) concerns schema theory as a possible explanation for how reading comprehension happens. This theory seeks to explain how old knowledge and new knowledge are related in the brain while reading. Schema help readers make inferences about how ideas are connected to each other by describing how the ideas relate to each other in temporal, causal, spatial, part-whole, and member-set ways. In other words, there are many kinds of ways to connect ideas to each other. As readers engage with a text, they connect these old and new ideas through making inferences. The four types of inferences that are made include deciding what schema to activate, holding in mind slots for the information in the schema, assigning default values for a schema unless conflicting information exists, and drawing conclusions to fill in any missing information. So, readers add to their knowledge of a topic by assimilating it into an existing, related schema or modifying existing schema where ideas may be in conflict. When this newly learned information needs to be recalled at a later date, it is accessed in one of three ways (i.e., there are three hypotheses for how schema theory explains memory). These hypotheses are the retrieval-plan hypothesis, the output-editing hypothesis, and the reconstruction hypothesis. Each of these hypotheses describes how the brain searches for information to link up a current schema to a relevant situation. Although schema theory may appear to be convincing, there is not enough evidence of how students make use of them when reading and remembering texts.

Walter Kintsch (1988) wrote the germinal piece on the construction-integration model. This model is a bottom-up response to the top-down processing proposed by schema theory. Although the model does not explain how inferences are generated, it does explain how these inferences are combined, or integrated, to make sense of a text. As readers engage with a text, they interpret its microstructure (i.e., the connections between words and sentences and entire paragraphs) as well as its macrostructure (i.e., the connection between paragraphs and the full text; the text’s argument, essentially). The combination of the microstructure and the macrostructure is the integrated with the reader’s prior knowledge of the topic to create what Kintsch calls the situation model. There is evidence from time-order activation analysis of when each type of information (e.g., microstructure, macrostructure, situation model) is activated while reading. This sequence of activation does not explain how the integration of inferences works, but it does describe the process of construction and integration as part of comprehension.

In 2005, Kintsch & Katherine A. Rawson further refined Kintsch’s original construction-integration model. In this refined model, there are four levels of comprehension: linguistic, microstructure, macrostructure, and situation model. The first two levels refer to words & phrases, and propositions, respectively. The third level, the macrostructure, refers to the connection between the propositions. The combination of the microstructure and the macrostructure is known as the textbase. The fourth level, the situation model, is the combination of the textbase and the reader’s prior knowledge and goals or purpose for reading. This model of comprehension is coherent because it can explain both on-line and off-line processing. On-line processing involves the integration of the textbase and prior knowledge, while off-line processing is the reconstruction of the inferences that were made to create the textbase. If you’ve ever stepped away from a text and felt that you understood it better upon reflection or the passing of time, then you have experienced the off-line processing discussed here.

Kintsch and Rawson have empirical evidence that shows readers making these types of inferences, which suggests the model is more robust than Kintsch’s original formulation. For example, readers pay more attention to the first sentences of paragraphs than they do later sentences in the paragraphs. This focus of attention contributes to the construction of the macrostructure. And, because conventionally structured writing often features important information in the first sentence of each paragraph, the reader can use that piece of prior knowledge to construct the situation model while reading. It follows from this attention to a text’s macrostructure that authors could help their readers learn new words from context by putting the most important words or concepts in the first sentence of each paragraph in a text, though this plan might not be feasible.

Before Kintsch and Rawson’s (2005) refinement of Kintsch’s 1988 model of construction-integration, Arthur C. Graesser, Murray Singer, and Tom Trabasso (1994) responded to the construction-integration model with the constructionist model. This model seeks to explain the different types of inferences readers can make. The authors list 13 types of inferences, but argue that not all are necessary. They describe certain conditions that need to be met for six of the inferences to be generated. The other seven types of inferences are said to be superfluous. As with the construction-integration model, there is empirical evidence supporting the constructionist model. This evidence comes in the form of preceding theory, verbal protocols, and time-order activation analysis. The verbal protocols present an additional source of information not captured in the construction-integration model, which gives the constructionist model additional explanatory power.

Incidentally, this article has one of my favorite tables in all of reading research (p. 375), which is a breakdown of the 13 types of inferences possible to make while reading Ambrose Bierce’s “How Leisure Came.” Here is that parable:

A Man to Whom Time was Money, and who was bolting his breakfast in order to catch a train, had leaned his newspaper against the sugar bowl and was reading as he ate. In his haste and abstraction he stuck a pickle-fork into his right eye, and on removing the fork the eye came with it. In buying spectacles the needless outlay for the right lens soon reduced him to poverty, and the Man to Whom Time was Money had to sustain life by fishing from the end of the wharf.

In particular, it is the 12th class of inference, involving “the emotion that the reader experiences when reading a text” that I find so endearing about this article. The authors cite the following text as eliciting the emotional inference: …on removing the fork the eye came with it. The inference made is simply stated as “The reader is disgusted.” Of course the reader is disgusted! Among all the other possible inferences to make when reading this text, the pure disgust it engenders is the most palpable response that I feel. Your mileage may vary, as the next model of reading comprehension indicates.

The landscape model of reading comprehension, published in 1999 by Paul van den Broek, Michael Young, Yuhtsuen Tzeng, and Tracy Linderholm, is a response to the construction-integration model and the constructionist model in that it attempts to explain individual differences in reading ability, which is something the construction-integration model does not do. The landscape model can also be thought of as a response to schema theory because it attempts to explain how different mental slots are activated or accessed during reading. The authors use co-activation to explain the role of prior knowledge; concepts that co-occurred in previous readings will be co-activated in the current reading. Thus, reading comprehension involves the interaction between the reader’s attention, the text, the reader’s background knowledge, and the reader’s ability to retrieve information. Where the idea of a landscape matters is that each reader has a different idea of what is important enough to activate prior knowledge about.

In other words, our lived experiences (which include reading) are varied, so certain words or concepts may “pop” for us when we see them, but they may not “pop” for others because those others have not had the same experiences. Each of our mental landscapes is different because of our varying background knowledge, so we will each take something different from the same text. Further, the co-activation of concepts in a reading cycle (i.e., a sentence) suggests the importance of context in learning vocabulary. If the context is rich with concepts related to the meaning of an unknown word, then the reader will co-activate these related concepts when the word is seen in a new context. This series of co-activations may explain how context is used to infer the meaning of unknown words, an additional explanatory benefit of the landscape model of reading comprehension.

If you take nothing else away from reading this post, it should be that the only skill or strategy involved in reading comprehension is inferencing. Connecting known information to new information in a text through making an inference is how we make meaning of text. If we do not know enough information about what we are reading, then we will not comprehend it. We will each take a different meaning from the text because we have different levels or types of background knowledge on any given topic. We can add to that knowledge through additional reading (or viewing or listening) and use that newly integrated knowledge to help us construct new meanings or deepen existing ones. When you are frustrated with someone not understanding your message online, consider that they have not had the same experiences as you have, so they don’t have the same associations between certain words or ideas. This idea seems so simple as to be tautologous. It’s kind of funny how the greatest contributor to reading comprehension is previous comprehension.

The Tall Rob Report

2025/04/08

van den Broek, P., Young, M., Tzeng, Y., & Linderholm, T. (1999). The landscape model of reading: Inferences and the online construction of a memory representation

No comments:

Post a Comment