Natural Language Parser

I gave a talk here on A Psycholinguistic Model of NLP implemented in Simulated Neurons in 2009.
The idea is that the semantics of the words are represented by neural firings derived from WordNet Synsets.
The semantics of the NP is represented by a frame, and the semantics of the VP and Sentence are represented by another case frame.
The frames support a memory based parser, so we can avoid a stack.
The word's lexical class is activated when a word is presented.
In collaboration with the existing frames, a grammar rule is selected, and this sticks the word into the appropriate frame.
The items are bound via short-term potentiation, so the frames can be reused later on.