Phrase-structure grammar (PS): Constituent Structure Analysis & Transformational Relations
1.3 PS grammar
The PS grammar, or phrase-structure grammar, is a model of generative grammar used to analyze the structure of different types of sentences in a particular language. The phrase-structure rules in this model indicate how a sentence may be divided into its constituents and how each constituent may be elaborated. The phrase marker, or P-marker, refers to the phrase structure of a sentence or its representation. A tree diagram can be used to exemplify the configuration of a sentence such as 'The man followed a girl.'
The points that are joined by the lines or branches are called 'nodes'. Each of the nodes, except those on the bottom line, is given a label that represents a grammatically definable constituent - N, V, NP, VP, etc. Where one node is higher in the tree than another and connected to it by branches, it is said to dominate it. If it is placed immediately above it and joined by a single line, it immediately dominates it. Thus, dominance shows how a larger constituent may consist of one or more constituents of a smaller kind.
However, phrase structure trees provide information of sentence structure, and the phrase structure grammar is able to state how new sentences can be generated. And the PS grammar does its operation through some rules, such as the following, to interpret the example sentence - 'The man followed a girl':
1- S → NP - VP.
2- VP → V - NP.
3- NP → DET - N.
4- V → followed
5- DET → the, a
6- N → man, girl
To interpret these rules, the first one says that a sentence (S) can consist of a noun phrase (NP) and a verb phrase (VP) in that sequence, and the second that a verb phrase (VP) can consist of a verb (V) and a following noun phrase (NP), and so on. These rules provide all the information given in the tree diagram. The first rule, for instance, is matched in the diagram by the S node immediately dominating NP and VP, and the second by the VP node dominating V and NP.
The rules mentioned above do not generate only the sentence 'The man followed a girl'. For since both 'the' and 'a' are shown as determiners and both 'man' and 'girl' as nouns, the rules permit us to permute the determiners in each determiner position and the two nouns in each noun position and, in fact, to generate no less than sixteen different sentences including the following ones:
- A girl followed the man.
- The girl followed a man.
- A man followed a man.
In a similar manner, phrase structure rules for a longer sentence would generate not only 'The young man with a paper followed a girl in a blue dress,' but equally 'A girl in a blue dress followed the young man with a paper,' if both 'the young man with a paper' and 'a girl in a blue dress' are noun phrases.
Since English, like many other languages, has intransitive verbs, it clearly shows and follows that VP is not always expanded into V - NP, but may be expanded into V alone, for example, 'The man smiled'.
(VP → V).
Instead of composing two distinct rules for the expansion of VP, a single rule can be written to indicate that the presence of NP is optional by placing it in round brackets.
VP → V (-NP).
So a VP can be substituted for V only, or V + NP. Nonetheless, the other expansions of a VP which we see here under use of ‘A’ for an ‘adjective’ and PP for ‘prepositional phrase’ are plenty.
- Rafit seems happy. (VP
→ V - A)
- The boy sat on
the floor. (VP → V - PP)
- The girl made Rafit
happy. (VP → V - NP - A)
- I gave the man a book. (VP →
V - NP - NP)
These different structures are, of course,
closely associated with the lexical verbs
that are used. We can combine all
the rules into one:
But this does no more than summarize
all the possible constituents of a VP.
It does not imply that any, or
all of them, will occur with a given
verb.
However, there is a very different sense in which elements are 'optional'. Other than the elements indicated in the previous rules, prepositional phrases can be used with any verb nearly without limits, making them almost freely insertable within sentences. Even more obviously, attributive adjectives freely occur before nouns. We have both ‘the boys’ and ‘the little boys’ to show that we need a rule:
NP → DET (-A) -NP.
In fact, we will even have to suggest that several adjectives may occur in this position as in the case of "naughty little English boys." It is reasonable to look for other, more general, evidence of constituency in order to acknowledge that the constituents have grammatical functions that do not only narrowly restrict expansion.
Firstly, only constituents may be coordinated
with conjunctions such as "and":
- The young man followed
the girl and the child.
- The young man followed
the girl and spoke to her.
In the first sentence, "the girl"
and "the child" are coordinated, while
in the second "followed the girl"
and "spoke to her" are coordinated,
both being VPs. This test allows us
to recognize different constituency in:
- He ran down the
hill.
- He ran down the
company.
- He ran down the
hill and down the valley.
- He ran down the
hill and the valley.
Secondly, only constituents may act as
the antecedents of certain pronouns:
- I saw the old woman and spoke
to her.
Nonetheless, pronouns seem to have sentences
also as their antecedents as in:
- A. Shoilee is coming tomorrow. B. I
don't believe it.
- Rafit said Shoilee is
coming tomorrow, but I don't believe it.
Thirdly, it is argued that only constituents
may be omitted as in:
- Rafit will see the
man, but Shoilee won't.
This suggests that "see the man" is a constituent. If this is so, the modal verb "will" does not form a constituent with the rest of the verbal elements, and the constituency of the verb phrase is not "will see" and "the man" as is shown in the tree diagram:
This is contrary to the traditional view, and indeed the common sense one, that the auxiliary verb is a modifier of the main verb.
However, constituent structure analysis, in terms of PS grammar, taken together with all the relevant information from the lexicon, allows us to generate a vast number of sentences. And if we allow recursion, an infinite number can be generated. Nonetheless, it will fail to deal with some characteristics of language such as discontinuity, where we need a model that either somehow allows for crossing branches, or else permits elements to be moved from one 'original' position to another.
A phrase structure grammar also fails to show how sentences that look similar may in fact be very different in some way or to show how sentences that look very different are, nevertheless, closely related. It is problems of this kind that may be handled in terms of transformational relations. In the earlier literature, the most important of these relations is that between active and passive sentences, for this is clearly something that a phrase structure grammar cannot account for. We could consider two contrasting pairs of sentences. First-.
- The boy kicks the ball.
- The boy kicked the ball.
The relation between these is easily handled by a phrase structure grammar. They differ only in that one has a present tense ending to the verb (-s), while the other has a past tense ending (-ed). Apart from that, they are identical in their structure, and their analyses would show simply there was a difference in the endings that occur with the lexical element 'kick', a difference associated with present and past tense. However, the situation is very different with –.
- The boy kicked the ball.
- The ball was kicked by the boy.
It is quite reasonable to argue that these two sentences are also closely related grammatically, but that the difference is not one of tense, but of active and passive voice. There is no obvious method of showing this in a phrase structure analysis. Thus, a phrase structure grammar is not capable of dealing with transformations of sentences and their ambiguities. It rather simply identifies the phrases and their constituents of which a particular sentence is composed.
References: