TRANSFORMATIONAL GENERATIVE GRAMMAR
4.1 Outline the phrase structure component
Various terms have been used to refer to Transformational Generative Grammar among them being: generative grammar, transformational generative grammar and transformational grammar. In the context of this course, we will adopt the term transformational generative grammar because it best reflects the content covered.
OVERVIEW OF TRANSFORMATIONAL GENERATIVE GRAMMAR
Transformational Generative Grammar is associated with Noam Chomsky, an American linguist of Jewish extraction based in Massachusetts Institute of Technology. He was born in (By the time of writing this module he is alive and still fine-tuning his theory of grammar). His initial academic interest was in maths and logic and this has influenced the way he formulates his linguistic theories, especially in the use of symbols in order to achieve precision in scientific description. He has had the greatest impact on the history of linguistics, particularly in the area of syntax where he has published prolifically. His first major publication is Syntactic Structures, which appeared in 1957. So different was it from the earlier models of grammar that it was hailed as a revolution. It broke new ground in a number of ways as we shall see. He managed this feat despite being a product of American structuralism.
To counter the weaknesses evident in traditional grammar and structuralism, Transformational Generative Grammar envisages a theory of language that does more than analyse already formed sentences. It seeks to explain the nature of a native speaker’s competence (inborn knowledge) that enables one to:
- produce and understand an infinite number of sentences
- recognise grammatical and ungrammatical sentences
- perceive the structure of sentences (its various constituents)
- recognise sentences that are paraphrases of each other. For example, “Put the light on” and “Put on the light”.
- recognise sentences that are For example, “ Diligent men and women always achieve their goals”. Who is diligent? Just the men or both men and women?
This grammar is exclusively concerned with the form of sentences and distinguishes between ‘grammatical’ and ‘meaningful’ sentences. Chomsky uses the following famous example to show the difference: ‘Colourless green ideas sleep furiously’. No doubt, you can perceive that this sentence is grammatical because it has an acceptable English structure (in line with point iii above). It is equivalent to a sentence such as ‘ Healthy small children play energetically’. Both have a subject in the form of an NP, a predicator and an adverbial (SVA structure). Contrast this with: ‘* Furiously sleep ideas green colourless’. You will agree that the sentence is ungrammatical for its structure is not possible in English. Turning to meaning, we find that “ Colourless green ideas sleep furiously”, though grammatical, is semantically unacceptable for it is nonsensical.
A basic assumption of Transformational Generative Grammar is that a sentence has two levels. The first is the deep structure. This is an abstract underlying level where the meaning of the sentence lies. This level was totally lacking in structuralism. The second is the surface structure which carries the pronunciation of the sentence and resembles any ordinary sentence. To illustrate the difference between the two levels, let us revisit the ambiguous sentence in (v) above. The sentence itself is a surface structure and since there is only one of it, we shall say that there is one surface structure. However, as we have already seen, it has two meanings/interpretations. This then means that it has two deep structures. Also consider the two sentences in (iv) above. Do you realise that because there are two of them, we have two surface structures but only one deep structure since the two have one meaning? Further examples would be: ‘Ngugi wrote the novel’ and ‘The novel was written by Ngugi.’ Once again, we have two surface structures but only one deep structure.
Of importance is the fact the deep structure and the surface one are produced by different types of rules. The deep structure is produced by phrase structure rules. After this, the second type of rules, called transformational rules, come in to change that deep structure into a surface structure. This implies that a transformational grammar has two main components: the phrase structure component and the transformational one. Will now look at these two components in turn.
THE PHRASE STRUCTURE COMPONENT
As already mentioned, phrase structure grammar (PSG) provides rules that can be used to generate a countless number of well-formed sentences most of which may never have been produced before. We must understand that these rules are not regulatory (like those of traditional grammar that dictate what you can or cannot say). Rather, they are constitutive, meaning that they tell us what constituents go into generating a sentence.
Phrase structure rules have two parts. These are: (1) the rewrite rules (2) the lexicon. We begin by looking at rewrite rules.
Rewrite rules are also called expansion rules. They are used to produce the basic (also called kernel) sentences. Such sentences include: active (as opposed to passive), statement (as opposed to question or imperative) and positive (as opposed to negative). What is interesting is that though these rules are few, they have the potential to generate an infinite number of actual and possible sentences. They are ordered systematically and applied stepwise.
A rewrite rule consists of a single symbol on the left, followed by an arrow and then one or more symbols to the right of that arrow. For example: NP -> N. The arrow shows that the left symbol is “ rewritten as” / “ expanded as” or “ consists of” the symbol(s) to the right. The ‘plus’ symbol means ‘followed by’. In its place, a comma-like (, ) symbol could be used . The symbols used are abbreviations such as S for sentence and NP for noun phrase. They are always written in capital letters. Thus, we see that rewrite rules show the linear order of elements in the deep structure of a sentence.
Since the starting point is the sentence, the initial symbol on the left is S. There are two ways in which linguists rewrite the symbol S.
- In the first one, it is assumed that a sentence consists of a subject (in the form of a noun phrase) and a predicate. The symbol for the predicate is simply The VP consists of the Aux (short for auxiliary), the main verb and the constituents that may come after the main verb such as object, complement and adverbial. Following this approach, S is rewritten as indicated below.
S -> NP + VP.
The Aux is therefore taken to be an integral part of the VP just as is the case in traditional and structural grammar.
- In the second one, which is more recent, it is assumed that a sentence consists of a subject in the form of a noun phrase, an Aux and a VP. The only difference between this approach and the one above is that here the Aux is regarded as a separate constituent from the VP (predicate), hence:
S-> NP + Aux + VP
There are two syntactic reasons given for this separation. To understand them, observe what is happening in the following sentences.
1a) Can he ride the bicycle?
1b) Yes, he can or Yes, he can do so. 1c) He can certainly ride the bicycle.
In (1b), there are two responses: In the first one, the modal auxiliary ‘can’ (which is part of Aux) occurs minus the rest of the predicate. In the second one, the auxiliary occurs together with a proform (do so) that stands for the rest of the predicate (ride the bicycle). In (1c), an adverb has separated the auxiliary from the rest of the predicate.
So, the two syntactic reasons are that the auxiliary can occur independently and that it be separated from the predicate by an adverb.
There is also a semantic reason for the separation. While the main verb denotes the state, action or event, the auxiliary elements express grammatical notions such as tense, aspect, voice and modality.
It is for these reasons that we will adopt the second approach in our presentation of rewrite rules and the attendant tree diagrams. Below, we present the rewrite rules.
- S-> NP + Aux + VP
At this point, we know how to rewrite a sentence but not an NP, an Aux or a VP. This leads us to rule 2 in which an NP can be written in various ways. Note that the only obligatory symbol is the N (or Pron if a pronoun is used). We will therefore put the others within brackets to show that they are optional.
- NP-> N g. sweets
(D/Det) + N e.g. (the) sweets
(D) + (AP -adjective phrase) + N e.g. (the) (red) sweets
(D) + (AP) + N + (PP) e.g. (the) (red) sweets (in my purse) PN (proper noun) e.g John
Pron (pronoun) e.g. They
Next, we rewrite the Aux . Note that the only element of Aux that is obligatory is tense; the others are optional. ‘M’ stands for modal, ‘have + en’ for the perfect aspect and ‘ be + ing’ for the progressive aspect.
- Aux -> Tense + (M) + (have + en) + (be + ing) g. could have been writing.
- Tense -> present past
We then move to the VP.
- VP-> V + (NP) + (PP) + (ADVP) e.g. ate (food) (in a hotel) (yesterday). Note that the verb could also be followed by an AP as in ‘It looked very good’. ‘Looked’ is a verb and ‘very good’ an adjective phrase
- PP -> P + NP e.g. on the desk
- AP -> (Adverb) + adj e.g. (quite) new
- ADVP-> (Adverb) + adverb e.g. (so) quickly
Perhaps we need to revisit the rule that rewrites the Aux to make it clearer, especially regarding ‘have + en’ and “be + ing’. Earlier, we said that phrase structure rules produce the deep structure and we know that this is where meaning is represented. With that in mind, let us consider the phrase ‘have eaten’ which expresses the perfect aspect. You might be surprised to learn that although ‘-en’ is suffixed to ‘eat’ to form ‘eaten’, it is really more closely related to‘have’. How so? The two work together to express the meaning that the action was completed prior to the moment of speech (perfect aspect). That is why we cannot say: ‘* I have eat or I eaten’. The point is that while the verb ‘eat’ expresses the action involved, ‘have’ and ‘-en’ jointly express the perfect aspect (hence have + en). Note that instead of ‘-en’, we could have ‘-ed’ as in ‘ have talk-ed’ or no suffix at all as in ‘have lost’. Thus, ‘-en’ is used as a cover term to refer to the perfect aspect morpheme.
Likewise, the meaning that the action is still continuing (progressive aspect) is carried by ‘ be + ing’. No wonder then, that we cannot say: ‘* I am eat or I eating’. Remember that in this particular case, ‘be’ could be realised in any of the following forms: is, was, are, were, am . ‘-ing’ is therefore more closely related to ‘be’ than it is to ‘eat’.
In the unit English Grammar and Usage, we learnt about auxiliaries, both primary (be, have, do) and modal (may, can, should, would and so on). Hopefully, you have realised that in PSG, Aux not only includes these two types of auxiliaries, but also tense, which is compulsory.
Recall that we said that phrase structure rules also include the lexicon. The lexicon is like a dictionary in that it consists of a list of words and affixes. The difference is that the lexicon is more detailed in giving information about the features of the words and affixes. This information relates to spelling, pronunciation, affixes associated with a word or bases associated with an affix, the syntactic environment in which a word can occur, and the meaning of the word or affix. Also, while the dictionary is a physical entity, the lexicon only exists in our minds and is therefore a mental dictionary. In other words, the knowledge that we have in our minds about words and affixes and their features, is way beyond what appears in any written dictionary. What then is the purpose of the lexicon in PSG? So far, the rewrite rules have merely given us symbols like S, NP and D. As you know, in speech we use words, not such symbols. This is where the lexicon comes in – to provide us with words. Supposing we have the following rewrite rule: S -> NP + Aux + VP where the NP consists of D + Adj + N. The Aux consists of tense + M + have+en.
The tense is present. VP in turn consists of V + NP. The NP consists of D + N. The whole string will now look like this: S-> D + Adj + N + present tense + M + have +en + V + D
Can you think of words that you can use to satisfy this rule? From the lexicon, we can select the following words to replace the symbols:
D – the , Adj – tall, N – stranger, present tense, M – may, have + en, V – abduct, D – a, N
This will give us the sentence:
The tall stranger may have abducted a child.
Further examples are:
- S-> Pro + past tense + modal + V + ADV Possible sentence: He could arrive today
- S -> PN + present tense + have+en + be +ing + V + P+ D+N Possible sentence: Peter has been walking on the
- S -> N + past tense + be+ing + V Possible example: Flowers were
- S -> D + Adj + N + P+ D+N + present tense + V Possible sentence: The young man at the corner
By now, you may be puzzled by what is happening to the elements of the Aux such as tense. Tense attaches to the verb on its immediate right. Therefore, in (1) above, past tense attaches to the modal ‘can’ to produce the past tense of that modal which is ‘could’. In (2) above, the present tense attaches to the primary auxiliary ‘have’ to produce the present 3rd person auxiliary verb which is ‘has’. In (3), the past tense attaches to the primary auxiliary ‘be’ to give us ‘were’
At this point, you should be in a position to provide a rewrite rule for a given sentence. Here are examples:
- The pupils should have greeted the
Rewrite rule: S-> D + N + past tense + M + have+en + V + D + N
- They fought
Rewrite rule: S-> Pro + past tense + V + ADV
- Some rare birds have been flying into
Rewrite rule: S-> D + Adj + N + present tense + have+en + be+ing + V + P + N
Both the symbols and the words can be presented in the form of branching tree diagrams. In PSG (and transformational generative grammar in general) such trees are referred to as phrase-markers.
Looking back on what we have said so far, we see that phrase structure rules do not generate finished sentences since no ordinary sentence would have ‘be+ing’ on the surface. Instead, these give us the underlying (deep) structure of sentences.
WEAKNESSES OF PHRASE STRUCTURE GRAMMAR
Among the shortcomings of PSG are the following:
- It is not able to generate some types of sentences since it only caters for the kernel ones. The complex ones are therefore not addressed. Here is one such example: ‘ The chief said that the thief who stole the cow believed that he had not been seen’.
- It does not capture the relationship between sentence types as in the case of active versus passive
- It is not able to handle discontinuous elements such as the auxiliary and the main verb in In ‘Are you eating with us ?’ for example, ‘are’ and ‘eating’ have been separated by the subject ‘you’ yet the two of them are part of the VP.
- Cannot handle stylistic variations such as ‘ Miraculously, the pilot survived the plane crash / The pilot miraculously survived the plane crash / The pilot survived the plane crash miraculously. Instead of capturing the fact that the three versions are paraphrases of one another, PSG gives a different rewrite for each.
- Much as it could handle structural ambiguity (occasioned by the arrangement of words), it had no answer for lexical ambiguity such as ‘I found it at the bank’. Are we referring to the edge of the river or a financial institution?