I think it's really cool how much I can relate programming knowledge to the linguistics course I'm taking.
A slight background first - trees. Trees are cool. In programming, trees allow for quick order-based searching for just about any amount of data. Take whatever, and just put it in some order. I've already gone through several instances where trees can be used in past posts (like my prefix/infix/postfix post or my regex post) so I won't go in too much here. However, I do appreciate the creation of the binary search tree, where everything on the left branch must be smaller and everything on the right must be larger than the node value. Makes searching real easy.
In linguistics, instead, there is a definition for how to form a sentence in a binary tree (like) format. We define any part of a sentence as a "phrase". Thus "dog" is a noun phrase. "eat" is a verb phrase.
A phrase X has two components - a specifier, and the main phrase part (X'). The main phrase part also has two components, the head (the main part), and the complement (or continuation). For example, in the noun phrase "a dog", "a" is a specifier, and "dog" is the head of the phrase in the main phrase part. If we wanted to continue this, we could say "the dog with the bone", thus: noun phrase with preposition phrase complement, preposition phrase has a noun phrase complement.
This is sorted into trees. Using notation of Value(Leftbranch, Rightbranch), "the dog with the bone" would be: NP(the, N'(dog, PP(P'(with, NP(the, N'(bone)))))). Now I have a phrase in binary tree form. It's also cool how the binary search tree analogy works here. The words that follow have a larger "value" than those preceding them due to their position in the phrase. Thus searching just follows the right branch downward until the right position is found.
There are more complicated concepts like moving wh- words, or inserting do into questions, but everything still manages to follow the same binary tree format like in programming. How interesting.
No comments:
Post a Comment