🚀 go-pugleaf

I would like to know whether has anybody tried to
seriously apply the idea of mental models (I refer
to the idea outlined by Johnson Laird in his well-known
book 'Mental Models') to text-analysis. I started a
research in this direction few months ago and I think
it'd be useful to compare my results with the ideas of
other students interested in this subject.

It's extremely important to define a linguistic framework
in which you can identify the objects that have to
correspond to a specified kind of token in the mental
model; that is, to define a taxonomy of the possible
entities men are able to address while speaking. Of
course this is very problematic and I think there is no
chance to get to a complete understanding of language
within a single framework. Any framework must then try
to define a view on this problem and work inside it.

I(try to) use a 'natural' taxonomy for entities based
on a semantic distinction. All the entities of the
language get classified under one of the seven following
choices, according to their functionality:

WHO, WHAT, WHERE, WHEN, HOW, WHY, and HOW MUCH (this
last one might be unnecessary).

For example the following sentence (from International
Business Week, september 7, 1992, page 29):

'The country is moving ahead with economic reform
despite the deepening troubles of President Fernando
Collor de Mello.'

would get analysed as the following:
 
  The country -> WHAT
  is moving -> (not classified at this stage)
  ahead -> HOW
  with -> (not classified at this stage)
  economic reform -> WHAT
  despite -> (not classified at this stage)
  the deepening troubles of Pres. F. Collor de Mello -> WHAT

What's the meaning of this? The meaning of this
classification is that you 'll have a token for
every WHAT and WHO, and you'll have a relationship
(spatial or temporal) for every WHERE and WHEN found.
HOW and HOW MUCH are used to modify or specify verbs
(as in the example above). WHY is been useful so far
only when you get to analyze complex sentences.
Prepositions and connectives are useful to identify
the functions of the parts of speech.


If we'll be able to find a principle to classify
the parts of speech in this way (or in a similar way)
and to combine the classification with an analysis
of verbs and an analysis of what is inside every single
token-description, we probably make a step forward in
explaining how we comprehend texts and discourses.


Whoever is interested in this approach at the
comprehension of texts, or is able to give me
suggestions or information on bibliographical
sources, is welcome.

Thanx.