Foltz
Peter W. Foltz in Hypertext and Cognition
(1996)
User models
Kintsch model
Levels of processing
individual words - meaning or gist of text
surface representation of words and sentences
textbase - meaning of text
situation model-general representation incorporating reader's background knowledge
Propositions are semantic primitives for information acquired
Coherence
propositions are connected via semantic coherence
semantic coherence depends on shared meaning or reference between constituents of text
coherence achieved by use of rhetorical devices: causality, pronouns, word repetition.
Reader generates a macrostructure, based on hypotheses and inferences about the overall meaning of the text, which connects to subordinate representations of lower level propositions.
Bridging inferences
when coherence between propositions is not present in the form of shared arguments, the reader must make a bridging inference, using backgound knowledge to fill in the missing information.
If the relevant background knowledge is not there, the bridging inference can cause lower comprehension or total confusion.
In a linear text, much coherence is provided by the writer - references which can't be assumed to be in the reader's background knowledge are explained before presenting propositions which depend on them.
In hypertext, the reader may encounter a chunk of text from many different starting points, and may not have touched on some vital fact or concept.
Navigating the hypertext will pose additional problems for the reader with
little background knowledge as it will not be clear which path will lead to
the desired information.
Hypertext
On the web, the reader may be linking in from another site altogether.