Figure 1210

Memory for surface and situation representations for stories described as literary or as newspaper reports. report perspective Data from Zwaan (1994).

flexible across different situations. In contrast, the weak rules incorporated into the construction-integration model are much more robust, and can be used in virtually all situations.

On the basis of a fairly detailed specification of the model, Kintsch (1988) was able to produce some convincing computer simulations of parts of the model. Further testing of the model was carried out by Kintsch et al. (1990). They tested the assumption that text processing produces three levels of representation ranging from the surface level based directly on the text itself, through the propositional level, to the situational or mental model level (providing a representation that is similar to the one that would result from directly experiencing the situation described in the text). Participants were presented with brief descriptions of very stereotyped situations (e.g., going to see a film), and then their recognition memory was tested immediately or at times ranging up to four days.

The main findings are shown in Figure 12.9. The forgetting functions for the surface, propositional, and situational representations were distinctively different. There was rapid and complete forgetting of the surface representation, whereas information from the situational representation showed no forgetting over four days. Propositional information differed from situational information in that there was forgetting over time, and it differed from surface information in that there was only partial forgetting. As Kintsch et al. (1990) had predicted, the most complete representation of the meaning of the text (i.e., the situational representation) was best remembered, and the least complete representation (i.e., the surface representation) was the worst remembered.

Zwaan (1994) tested the psychological reality of some of the levels of representation identified in the construction-integration model. He argued that the reader's goals influence the extent to which different representational levels are constructed. For example, someone reading an excerpt from a novel might be expected to focus on the text itself (e.g., the wording; stylistic devices), and so form a strong surface representation. In contrast, someone reading a newspaper article may focus on updating his or her representation of a real-world situation, and so form a strong situation representation. Zwaan (1994) devised texts that were described as literary extracts or news stories. As predicted, memory for surface representations was better for stories described as literary, whereas memory for situation representations was better for stories described as newspaper reports (see Figure 12.10).

The reality of the distinction between textbase and situation representations was studied by McNamara, Kintsch, Songer, and Kintsch (1995). They presented participants who had high or low knowledge about the heart with a technical text about the heart's functioning. There were two versions of the text, one of which was more coherent and easier to read than the other. It would be expected that construction of the textbase and of the situation model would both be facilitated with the more coherent text. This was the case for low-knowledge readers. In contrast, high-knowledge readers showed slightly increased textbase recall with the more coherent text, but constructed a less effective situation model.

What do these findings mean? They suggest that knowledgeable readers benefit from a somewhat incoherent text that makes them fill in the gaps and prevents them from engaging in superficial processing. At a theoretical level, the notion that there are separate textbase and situation representations is supported by the finding that text coherence had opposite effects on these two representations for high-knowledge readers.


One of the greatest advantages of the construction-integration model over the Kintsch and van Dijk (1978) theory is the assumption that there is a situational representation or mental model as well as a propositional representation. An interesting study suggesting the superiority of the mental or situation model approach to the propositional approach was conducted by Glenberg, Meyer, and Linden (1987). They presented short passages such as the following:

John was preparing for a marathon in August. After doing a few warm-up exercises, he (took off/put on) his sweatshirt and went jogging. He jogged halfway round the lake without too much difficulty.

Further along the route, however, his muscles began to ache.

The participants were then probed with the word "sweatshirt", and had to decide as rapidly as possible whether it had occurred in the passage. They responded significantly faster with the version of the story in which John put on his sweatshirt than when he took it off. This finding is rather mystifying from the perspective of a propositional theory, because the probe word is represented in only one proposition in both versions of the passage. In contrast, the finding is readily explicable from the mental or situation model approach: if participants construct mental models depicting the events described in the passage, then the sweatshirt has much greater prominence when John puts it on than when he takes it off.

The construction-integration model has the advantage over its predecessors that the ways in which information in the text combines with related knowledge already possessed by the reader are spelled out in much more detail. In particular, the notion that propositions for the text representation are selected on the basis of a spreading activation process operating on propositions drawn from the text and from stored knowledge is an interesting one. This approach seems superior to the previous model's emphasis on integration based on linking together propositions sharing an argument.

Situational representations are not always constructed. Zwaan and van Oostendop (1993) asked their participants to read part of an edited mystery novel describing the details of a murder scene, including the locations of the body and various clues. Most participants did not construct a situational or spatial representation when they read normally. However, such representations were constructed at the cost of a marked increase in reading time when the initial instructions emphasised the importance of constructing a spatial representation. These findings suggest that limited processing capacity may often restrict the formation of situational representations or mental models.

According to Graesser et al. (1997), there are two levels of discourse representation in addition to the three levels identified by Kintsch (1988). There is the text genre level, which is concerned with the nature of the text. Text genres include narration, description, jokes, exposition, and persuasion. The kinds of information presented, how the information is presented, and the ways in which the information is to be interpreted differ greatly across genres. There is also the communication level, which refers to the ways in which the writer tries to communicate with his or her readers. However, some readers may not form a representation at the communication level. As Graesser et al. (1997, p. 169) pointed out, "The reader of a novel may not construct an invisible, virtual writer or storyteller that communicates with the reader, unless there are explicit features in the text that signal that communication level."

There are two other major problems with the construction-integration model. First, it is as sumed within the model that numerous inferences are considered initially, with most of them being discarded before the reader becomes aware of them. This key theoretical assumption has not been tested systematically. Second, the model as a whole has not yet been put to a searching test. Evidence supporting parts of the model has been obtained, but this evidence is insufficiently detailed to be convincing.

Event-indexing model

The construction-integration model is not specific about the processes involved in the construction of situation models. However, Zwaan, Langston, and Graesser (1995) put forward an event-indexing model to remedy this omission. According to this model, which was further developed by Zwaan and Radvansky (1998), readers monitor five aspects or indexes of the evolving situation model at the same time when they read stories:

1. The protagonist: the central character or actor in the present event compared to the previous event.

2. Temporality: the relationship between the times at which the present and previous events occurred.

3. Causality: the causal relationship of the current event to the previous event.

4. Spatiality: the relationship between the spatial setting of the current event and that of the previous event.

5. Intentionality: the relationship between the character's goals and the present event.

Key predictions of the event-indexing model are that discontinuity in any of these aspects (e.g., a change in the spatial setting; a flashback in time) creates difficulties in situation-model construction, and that discontinuities increase reading times for events. Zwaan and Radvansky (1998) referred to an unpublished study of theirs in which the participants rated how well each sentence fitted in with the previous sentences. These ratings were determined by the overlap between the five situational dimensions in the target sentence versus the previous sentences: the greater the overlap, the higher were the ratings.

Additional support for the event-indexing model was reported by Zwaan et al. (1995). They found the reading times for events in a story increased as a function of the number out of the five situational dimensions on which there was discontinuity with the previous event. They also showed that each of the five dimensions had its own influence on reading time.


The greatest strength of the event-indexing model is that it identifies some of the most important processes involved in creating and updating situation models. The emphasis of the model on the construction of situation models is probably well placed. As Zwaan and Radvansky (1998, p. 177) argued, "Language can be regarded as a set of processing instructions on how to construct a mental representation of the described situation."

On the negative side, as Zwaan and Radvansky (1998, p. 180) admitted, is the fact that the event-indexing model, "treats the individual dimensions as independent entities." This approach is unlikely to be correct, because the various situational dimensions seem to interact in various ways. Consider the following sentence provided by Zwaan and Radvansky (1998): "Someone was making noise in the backyard; Mike had left hours ago." This sentence provides information about temporality. However, it has relevance to the causality issue, because it permits the causal inference that Mike was not the person making the noise.

• Sentence processing. Sentence processing involves parsing and the assignment of meaning. According to the garden-path model, only one grammatical structure is initially considered, and meaning does not influence the selection of this structure. According to content-guided processing theory, meaning does play a role in determining the assignment of syntactical structure. According to constraint-based theory, various processes operating in parallel use the constraints imposed by relevant semantic and syntactic information to favour one particular syntactic structure. A limitation of this theory is that many of the details of syntactic processing are not specified. Inner speech is used to preserve information about word order. It may also be used to provide prosodic structure.

• Capacity theory. According to Just and Carpenter's capacity theory, there are individual differences in the capacity of working memory. These individual differences have substantial effects on language comprehension. Working-memory capacity is assessed by the reading-span task. There is evidence based on fMRI to indicate that the various processes involved in the reading-span task use the same areas of the brain as are used in sentence comprehension. It has been argued that what is common to all aphasic patients is a reduction in the working memory resources available for language comprehension. Capacity theory tends to ignore the various specific processas involved in comprehension.

• Discourse processing. According to the minimalist hypothesis, only a few inferences are drawn automatically; additional strategic inferences depend on the reader's goals. This contrasts with the constructionist viewpoint, according to which numerous automatic inferences are drawn. A reasonable compromise is provided by the search-after-meaning theory. This assumes that readers try to construct coherent meaning for texts based on their goals, and they try to explain actions and events in those texts.

• Story processing. According to schema theory, schemas or organised packets of knowledge help to determine what we remember of stories. Recall of texts often includes schematic information that was not presented. Schemas influence comprehension and retrieval processes. There is some support for a specific form of schema theory known as the script-pointer-plus-tag hypothesis. According to Kintsch and van Dijk's model, texts are processed to produce a micro-structure and a macro-structure. According to Kintsch's construction-integration model, three levels of representation of a text are constructed: the surface representation; the propositional representation or textbase; and the situation representation. The surface representation is forgotten most rapidly and the situational representation most slowly. The processes involved in the formation of situation models were identified in the event-indexing model. According to this model, readers monitor five aspects of the evolving situation model: the protagonist; temporality; causality; spatiality; and intentionality. Discontinuity to any of these aspects creates difficulties in situationmodel construction and increases reading times.

Business Correspondence

Business Correspondence

24 chapters on preparing to write the letter and finding the proper viewpoint how to open the letter, present the proposition convincingly, make an effective close how to acquire a forceful style and inject originality how to adapt selling appeal to different prospects and get orders by letter proved principles and practical schemes illustrated by extracts from 217 actual letter.

Get My Free Ebook

Post a comment