This is a fantastic challenge for computer engineers: create an interpreter or compiler for a programming language, based on statistic modeling. How likely is i

This is a fantastic challenge for computer engineers: create an interpreter or compiler for a programming language, based on statistic modeling. How likely is it that combining the token "foo" with "bar" using a mathematical operator is a valid construct, based on experience from other programs? What will the resulting value usually be, based on prior experience?