A theory of grammar that could produce a linguistically competent machine
The main assumption is pattern recognition.
I've described pattern recognition as the ability to determine the statistical impracticality of randomness. Once I can find out how to do this, I will take it from there.
My question to a coder would be this. How can I record every time that a random string of symbols occurs more than once and save that as a statistic?