GCL can be used to build symbolic models that are associated to external reality, e.g. the positive sciences.
It is supposed that GCL occurs by spontaneous interaction between people but this is an abnormal mode of occurence. It is not clear to me that a language can start from scratch, but let's suppose so. The abnormal mode of occurence is associated with another aspect. The language for any device used to process information (as the brain is) is made of a collection of terms and relations. Any element/relation of the symbolic model (language) must be associated with a component/function of the hardware. That is, the hardware must be known before the language is built. This is the normal situation, e.g. when a computer, which has no associated programs, has to be used.
But, as it is believed, the language used by the brain appeared without knowing the hardware. The main consequence is that all the words, which have to be associated with the basic feature of the hardware, have no precise definition.
Thus, we find in dictionaries what I call "external definition" of the words.
That is, such definitions are not based on the hardware. MDT as a theory
associated with the hardware, generates "internal definitions" of the words.
Some such definitions will be given below.
Dictionary of internal definitions for some words:
1. To believe: there is an incomplete (unstable) model. Such a model could become stable (harmonic/logic) if some artificial elements/relations are included (artificial means that something is not generated by the interaction with the external reality). After such changes, the model becomes stable. Any truth generated by such a model must be associated with the word "to believe". Also, the artificial changes must be specified before.
2. To know: there is a stable model (harmonic/logic) which is integrated in a stable structure of models. Any truth generated by such a model can be associated with "to know". I want to emphasize that from this does not result in any way that the truth is correct, when compared to the external reality. "To know" means just that the whole structure of models of the brain supports that truth, and nothing more! As one can see, "to know" is associated only to the structure of models, and not to the external reality.
3. As I know: there are some models which support a truth but some other related models are not good enough to support that truth.
4. Correct, to be correct: this term has at least two meanings. 4a. There is a model generating a prediction in association with the external reality. This prediction is compared with IR. If the result is positive, then the truth is correct. 4b. There is a stable structure of models. Such a structure has already predicted a large number of correct (4a) truths. In such a situation, any truth generated by the structure is considered to be correct (see also the definition of the term "to know").
To be correct based on definition 4a means to make an experiment (any comparison between a prediction and IR is called "experiment"). There are a very limited situation when an experiment can or may be done (e.g. if the problem is to verify if a bridge will survive or not in case of an earthquake, then such a problem cannot be solved based on an experiment).