Transformational generative grammar

Transformational generative grammar

In 1957 Noam Chomsky, an American, published Syntactic Structures, a statement of the principles of transformational generative grammar (TG).

 

This grammar has had a profound effect on the study of all languages, including English. TG was a reaction against structuralism and the first model to acknowledge formally the significance of deep structure.

 

Transformational generative grammarians set themselves the task of creating an explicit model of what an ideal speaker of the language intuitively knows. Their model must assign a structure, therefore. To all the sentences of the language concerned and only to these sentences.

 

As a first step towards this, Chomsky distinguished between 'competence', which he defines as 'the ideal speaker-hearer's knowledge of his language', and 'performance', which is 'the actual use of language in concrete situations'. Competence is, as it were, the perfect storehouse of linguistic knowledge. Performance draws on this knowledge but it can be faulty.

The TG model attempts to formulate hypotheses about competence by idealising performance, that is, by dredging away performance accidents such as hesitations, unnecessary repetition, lack of attention, fatigue, slips of the tongue, false starts. TG is interested in competence and this interest marks the clearest difference between structuralism and TG. Structuralism was text-based and only interested in language that had actually occurred. TG does not use text since it is more interested in what produced the text than in the text itself.

 

A TG model has four main characteristics:

 

1. It must attempt to make explicit how a finite entity like the brain can operate on a finite set of items (words and structures) and yet generate an infinite set of sentences. The model must parallel the ideal speaker's competence and so it must be capable of generating an infinite set of sentences by the operation of a finite set of rules on a finite set of items. We can give an impression here of how that can be done. Let us suppose, for example, that we have the rules:

 

 

S — NP + VP (sentence can be rewritten as noun phrase + verb phrase)

NP —  (det) + N (noun phrase can be rewritten as (determiner) + noun)

VP ~ V + NP (verb phrase can be rewritten as verb + noun phrase)

 

 

and suppose we have two nouns 'boys' and 'girls', three determiners 'the', 'some' and 'five', and three verbs 'love', 'hate' and 'trust', then we can produce hundreds of sentences such as:

 

• Boys love/hate/trust girls.
• Girls love/hate/trust boys. 
• Some boys love/hate/trust girls.
• Boys love/hate/trust some girls.
• Five boys love/hale/trust the girls.
• The boys love/hate/trust some/five/the girls.

These sentences give a limited idea of the productive quality of even the simplest model.

 

A TG model has four main characteristics:

 

2. Since the model attempts to describe the idea (speaker-hearer's linguistic knowledge and intuitions), it must be explicit. It must not fall back on intuition to ask whether a structure is or is not correct. If it used intuition to define intuition, the model would be circular and useless. A TG model must therefore be explicit and self-sufficient. Its rules alone must allow us to decide whether a structure is acceptable.

 

A TG model has four main characteristics:

 

3. The model must have three components:

• a phonological component,
• a syntactic component,
• a semantic component,

so that it parallels the speaker's ability to associate noise and meaning.

 

 

The phonological component deals with phonemes and with the permissible combination of phonemes. As far as English is concerned, it offers rules for stress and intonation patterns as well. The work on phonology is an extension of the work done by structuralists, a refinement rather than a reappraisal, and this is the part of the TG model which has received least criticism.

The semantic component deals with meaning and the interpretation of meaning. Much work has been done in this area and many have criticised Chomsky's techniques. It would be true to say, however, that less satisfactory work has been done with regard to semantics than with regard to phonology and syntax.

 

A TG model has four main characteristics:

 

4. Although the model must not rely on the intuition of a native speaker it must be in harmony with such intuition. In other words, it must be able to assign a structure to all sentences which would be accepted by a native speaker and reject all sentences which would be rejected by a native speaker.

 

It is with regard to his treatment of syntax that N.Chomsky's approach differs most fundamentally from other models.

 

 

TG is explicit about the fact that native speakers recognize two levels of structure.

A speaker realizes that:

• John is easy to please. 
• John is eager to please.

may look alike but are different at some level in that the first implies:

• Someone pleases John .

and the second:

• John pleases someone .

Similarly, a native speaker recognizes that although:

• John loves Mary

looks very different from:

• Mary is loved by John

they are fundamentally very similar.

To account for the two levels that a speaker intuitively recognises, a TG model splits the syntactic component into two pans: a base subcomponent and a transforma­tional subcomponent.

 

The base subcomponent generates (that is, assigns a structure to) the deep underlying pattern so that we can represent it by means of a tree diagram (also called a 'labelled bracketing' and a 'phrase marker'), thus:

S    NP + VP

NP    det + N        

VP   V+NP                                         

The transformational subcomponent works on a phrase marker and so generates a surface structure. Again, a brief example may help. The structure:

det + N + V + det + N

underlies thousands of transitive sentences such as:

The cat swallowed the mouse.

The transformational subcomponent accounts for the transformation of such a sentence into such variants as:

• The mouse was swallowed by the cat.
• The mouse was swallowed.
• The swallowing of the mouse (by the cat)

and:

 

Transformation rules allow the grammarian to explain:

(1) deletion, for example   A+B+C >A+B:

• John ran away and Mary ran away ---- John and Mary ran away

(2) addition/insertion, for example, A+B >A+B+C:

• Go away --- You go away
• He has come ---- He has just come

(3) permutation, for example, A+B+C > A+C+B:

• Call John up --- Call up John

(4) substitution, for example, A+B+C >A+D+C:

• John arrived and Peter went in ---- On John's arrival Peter went in

In brief,

• TG grammar aims to pair a given string of noises with a given meaning by means of a syntactic component.
• TG model is neutral with regard to production and reception.
• The ultimate aim of TG is the understanding of language, of the universals common to all languages, and through this an understanding of the human mind.