After a discussion at MIT on linguistics, Peter Norvig wrote this very interesting note that started as a brief note but turned into an interesting essay.
I think that Norvig is about right.
He gives Chomsky some credit.
Norvig seems genuinely interested in examining Chomsky’s ideas.
I think that 50 years from now that programs that ‘understand English’ will have modules that you can credit to Chomsky.
They will have others that are semantic.
The latter will need a large data base but perhaps not be large themselves.
The biggest modules will perhaps be metaphorical.
See “colorless green ideas sleep furiously” illustrated!!
The following text is especially interesting.
Quote:
“And yes, it seems clear that an adult speaker of English does know billions of facts (for example, that one says “big game” rather than “large game” when talking about an important football game).
These facts must somehow be encoded in the brain.”
Someone once told me how to remember whether a verb takes “or” or “er” when forming a noun.
I remember that it seemed accurate and easy to remember.
Alas I forgot the rule.
I think, however the rule resides in my unconscious language areas.
Perhaps there are such rules for “big” vs. “large”.
(2016 It turns out that there are categories of adjective which are not entirely arbitrary.
The categories have a simple order which determines the order of adjectives.)
At this point in the article Norvig veers into an interesting discussion of philosophy of science and provides more quotes from famous scientists than I had seen.
Norvig:
Chomsky takes the opposite approach: he prefers to keep a simple, elegant model, and give up on the idea that the model will represent the data well.
Instead, he declares that what he calls performance data—what people actually do—is off limits to linguistics; what really matters is competence—what he imagines that they should do.
I would give Chomsky a bit more here.
The rules Chomsky loves are probably rules most of us speakers love too, but often fail to obey.
That we aspire to them is a phenomenon that needs explanation.
That explanation may not help to produce programs that understand English however.
I think sometimes the path to understanding some sentence is to modify it or the rules, or the word categories, to achieve a conformance.
Thus to understand “it quaked her stomach” may require accepting “quaked” as a verb.
We probably do this unconsciously and perhaps the English program must do it explicitly.
This explicitly requires grammar and statistics.
Overall I think Norvig wins his points but still I retain some reason to applaud Chomsky’s efforts.
The biological eye really seems to be seeking the ‘ideal camera’ that was mathematically understood at least by Descartes.
I think that there may be such a mathematical ideal and that formal logic is a partial model of that.
I think this is important to pursue, even if it does not help to build the English program.
Just as the theory of the camera helps us transcend the biological eye, a theory of language may help us transcend the biological mind.
I don’t think that this is Chomsky’s goal however.