Phrase Structure Rules

Phrase structure rules are ways of describing a given syntax language and are closely associated with the early stages of transformational grammar .They are used to break sentences of natural language into their “constituent parts” (also known as syntactic categories ), ie phrasal categories and lexical categories (also known as grammatical class ). A grammar that uses phrasal structure rules is a type of sentence grammatical structure – except in Computer science , which is only known as grammar , is generally context-free . Phrase structure rules as they are commonly employed operate according to the constituent relation and the grammar that employs phrasal structure rules is therefore a “constituent grammar” and as such is in contrast to the grammar of dependency , which are based on a dependency relationship.

Definition 

Phrase structure rules are generally of the following format:

{\ displaystyle A \ to B \ quad C}

Meaning that constituent {\ displaystyle A} is separated into two sub-constituents {\ displaystyle B} and {\ displaystyle C}. Some other examples of English to follow:

{\ displaystyle S \ to NP \ quad VP}

{\ displaystyle NP \ to Det \ quad N1}

{\ displaystyle N1 \ to (AP) \ quad N1 \ quad (PP)}

In the first rule we have: S ( sentence ) consists of an NP ( noun phrase ) followed by a VP ( sentence ). The second rule is: A noun phrase consisting of a Det (determinant) followed by an N (name). Some other categories: AP ( adjective clause ), AdvP ( adverbial clause ), PP ( prepositional phrase ), etc. Applying the rules of phrasal structures in a neutral way, it is possible to generate several sentences, in this case in English. But it is also possible that the rules generate syntactically correct, but semantically meaningless phrases. The following example demonstrates this:

Colorless green ideas sleep furiously

This sentence was constructed by Noam Chomsky as a demonstration that phrasal structure rules are capable of generating syntactically correct but semantically incorrect sentences. Phrase structure rules break the sentence into its constituent parts. These constituent parts are generally represented as tree structures. The tree for Chomsky‘s famous sentence can be processed as follows:

A constituent is any word or combination of words that is dominated by a single node. Thus each word is a constituent. In addition, the subject NP Colorless green ideas , and the smallest NP green ideas , and the VP sleep furiously are constituents. Phrase structure rules and tree structures that are associated with them are a form of immediate constituent analysis .

Top down 

An important aspect of sentence rule structure is that they see sentence structure from top to bottom. The category on the left side of the arrow is the largest constituent and the immediate constituents that are on the right side are smaller constituents. Constituents are broken successively into their parts as the list of rules for a given sentence follows. This top-down approach runs counter to much work done in modern theoretical syntax. In the Minimalist, for example, the sentence structure is generated from the bottom up. The joining process brings together small constituents and creates larger constituents until it reaches the largest, that is, the sentence. Theoretical syntax abandoned the rules of phrasal structures long ago, although its importance for Computational linguistics remained intact.

Alternative approaches 

Constituent vs. dependence 

Phrase structure rules as they are commonly employed result in a phrase structure view that is “constituent based”. Just as the grammars that employ these rules are constituent grammar (= grammatical sentence structure ), as opposed to dependency grammar , which views sentence structure as based on dependency.. Which means that in order for phrasal structure rules to be applied, one has to seek a constituent-based understanding of sentence structure. The constituent ratio is one to one or more correspondents. For each word in a sentence, there is at least one node in the syntactic structure that corresponds to that word. Unlike the dependency relationship, the relationship is one to one; For each word of the sentence, there is exactly one knot in the syntactic structure that corresponds to that word. The difference is illustrated in the trees below:

The constituent tree on the left can be generated by phrasal structure rules. Sentence S is increasingly broken down into smaller parts of constituents. The dependency tree on the right side could not be generated from the phrasal structure rules (at least not as they are commonly interpreted).