Cognitive grammar

I'd mentioned Langacker's cognitive grammar in this post, referencing here one of Langacker's introductory articles on the topic and a free Google book preview on the same topic.

On pp. 6-7 of the book Langacker says something about grammar that I've frequently said about categorization more generally. While there are overlaps between lexicon, morphology and syntax, that doesn't necessary indicate that they don't each have their own definitive boundaries. Traditional syntax, e.g., is defined with a boundary so impenetrable as to be completely unrelated to semantics. Whereas in CG the overlaps between these categories provides for how they relate and thereby opens such strict boundaries. It doesn't eliminate the boundaries but enriches and more accurately defines each domain.


On p. 10 this is reiterated in that Chomsky's generative grammar uses formal mathematical models, the latter which assumes that math itself is a self-contained abstraction with either Platonic essences, or Aristotelian categories with strict set theoretical boundaries, or both, at its base. This thread has given ample examples of this phenomena. Whereas CG is more along the connectionist and embodied lines.

Also of note is that in formal math the symbols are contentless, whereas for CG the symbols are indeed full of meaning (10). Looking at this previous post CG does have contentless objects called (image) schemas, comparing them to archetypes. The former requires no embodied substrate, the latter is an embodied substrate. Also recall Knox discussing image schema as archetypes here, here, here, here and some commentary here.