A lot of tasks which are based on language processing can only be explained by the usage of situation models. The so called situation model or mental model consists of five different dimensions, which refer to different sources. To comprehend a text or just a simple sentence, situation models are useful. Furthermore the comprehension and combination of several texts and sentences can be explained by that theory much better. In the following, some examples are listed why we really need situation models.
Integration of information across sentences is more than just understanding a set of sentences. For example:
“Gerhard Schroeder is in front of some journalists. Looking forward to new ideas is nothing special for the Ex-German chancellor. It is like in the good old days in 1971 when the leader of the Jusos was behind the polls and talked about changes.”
This example only makes sense to the reader if he is aware that “Gerhard Schroeder”, “Ex-German chancellor” and “the leader of the Jusos in 1971” is one and the same person. If we build up a situation model, in this example “Gerhard Schroeder” is our token. Every bit of information which comes up will be linked to this token, based on grammatical and world knowledge. The definite article in the second sentence refers to the individual in the first sentence. This is based on grammatical knowledge. Every definite article indicates a connection to an individual in a previous sentence. If there would be an indefinite article, we have to build a new token for a new individual. The third sentence is linked by domain knowledge to the token. It has to be known that “Gerhard Schroeder” was the leader of the Jusos in 1971. Otherwise the connection can only be guessed. We can see that an integrated situation model is needed to comprehend the connection between the three sentences.
The explanation of similarities in comprehension performances across modalities can only be done by the usage of situation models. If we read a newspaper article, watch a report on television or listen to a report on radio, we come up with a similar understanding of the same information, which is conveyed through different modalities. Thus we create a mental representation of the information or event. This mental representation does not depend on the modalities itself. Furthermore there is empirical evidence for this intuition. Baggett (1979) found out that students who saw a short film and students who heard a spoken version of the events in the short film finally produced a structurally similar recall protocol. There were differences in the protocols of the two groups but the differences were due to content aspects. Like the text version explicitly stated that a boy was on his way to school and in the movie this had to be inferred.
Situation models have a deep influence for effects of domain expertise on comprehension. In detail this means that person A, whose verbal skills are less than from person B, is able to outperform person B, if he has more knowledge of the topic domain. To give evidence for this intuition, there was a study by Schneider and Körkel (1989). They compared the recalls of “experts” and novices of a text about a soccer match. In the study were three different grades: 3rd, 5th and 7th. One important example in that experiment was that the 3rd grade soccer experts outperformed the 7th grade novices. The recall of units in the text was 54% by the 3rd grade experts and 42% by the 7th grade novices. The explanation is quite simple: The 3rd grade experts built up a situation model and used knowledge from their long-term memory (Ericsson & Kintsch, 1995). The 7th grade novices had just the text by which they can come up with a situation model. Some more studies show evidence for the theory that domain expertise may counteract with verbal ability, i.e. Fincher-Kiefer, Post, Greene & Voss, 1988 or Yekovich, Walker, Ogle & Thompson in 1990.
An other example why we need situation models is by trying to explain translation. Translating a sentence or a text from one language to another is not simply done by translating each word and building a new sentence structure until the sentence seems to be sound. If we have a look now at the example of a Dutch sentence:
Now we can conclude that the translation level between Dutch and English is not based on the lexical-semantic level; it is based on the situation level. In this example “don’t do something (action) before you haven’t done something else (another action)”. Other studies came up with findings that the ability to construct situation models during the translation is important for the translation skill (Zwann, Ericsson, Lally and Hill, in 1998).
People are able to learn about a domain from multiple documents. This phenomenon can be explained by a situation model, too. For example, we try to learn something about the “Cold War” we use different documents with information. The information in one document may be similar to other documents. Referents can be the same or special relationships in the “Cold War” can just be figured out by the usage of different documents. So what we are really doing by learning and reasoning is that we integrate information on the base of different documents into a common situation model, which has got an organized order of the information we’ve learned.
We have seen that we need situation models in different tasks of language processing, but situation models are not needed in all tasks of language processing. An example is proofreading. A proofreader checks every word for its correctness. This ability does not contain the ability to construct situation models. This task uses the resources of the long-term memory in which the correct writing of each word is stored. The procedure is like:
This is done word by word. It is unnecessary to create situation models in this task for language processing.