Readers' models of text structures: the case of academic articles
Readers' models of text structures: the case of academic articles 
This article proposes that the superstructure of text is important for the comprehension. A superstructer is, basically, the way a text is written, like a newspaper or a academic article. This theorem has been tested in two experiments.
In experiment one, the author asked twelve subjects to put together parts of an academic article that have been cut out and mixed up. These cut outs were text paragraphs, tables and figures. One test was run with and a second without the headings. After putting the text together a comprehesion test was done.
In experiment two, the author asked eight subjects to assign text paragraphs to four different sections: Intro, method, result and discussion (IMRD). One test was run on paper and a second one on a computer screen.
The results showed that the subjects were quite good in finding the right sections because they knew the superstructure for academic articles, even without really reading the text. This "where am I" knowledge seems important for comprehension and must be preserved when creating hypertext.
The article showed that the predictability of text is an important factor for comprehension.
Suggestions for further experiments.
More than 12 subjects for further experiments.
Explain why the inclusion/exclusion of the headings didn't make any difference.
What's the purpose of experiment two on a computer? It was the same text, the only navigation item a button to turn the pages.
Is it easier to put an article together when the heading is included?
What was the goal of the experiments?