Skill as Lexical Knowledge

Generalising from Language Acquisition

February 6, 2013

Noting how learning to program and learning a natural language resemble each other… that is, a Bayesian (iterative) learning model.

Example:

• First time your hear the terminology “pivot table”, you have no idea what it is, what it does, what it’s good for. And you go through some bland, technical (yet totally correct!) descriptions in a dry article, still have no idea what it is.
• Only when you go through a tangible case study, a hands-on tutorial of collapsing multiple data categories, you finally get a concrete sense of what it is.
• Then later when you hear people mention the terminology, your mind would automatically compare whether the usage and your understanding are accurate, then update, reconcile and consolidate your “Bayesian knowledge”.

If your hear someone say “map/reduce is like a pivot table” followed by multiple real-world examples of how map/reduce is used, your mind will automatically check the “goodness of fit” for the analogy, and dynamically updates the “Bayesian priors” for both “pivot table” (something you learnt before) and “map/reduce” (some term you are learning).

Semantic Transfer

February 6, 2013

This is exactly how textbooks become unreadable.

Yet, the strange thing is, once you read the humane version, then read the “change notes” showing the process of evolution bit by bit — the formal version suddenly feels totally comprehensible and no longer that off-putting.

Therefore, I imagine it’s possible to teach programming by “replaying” a complex project “diff-by-diff” and explaining each step.

See, we just reverse-engineered the process of making something understandable (in essence, learning).

EDIT: Note on “general learning is like learning a language”
cf. http://theconversation.com/keep-it-simple-stupid-maths-doesnt-have-to-be-complex-16909