THE SENTENCE PATTERNS OF LANGUAGE

THE SENTENCE PATTERNS OF LANGUAGE

4.1. Syntax as opposed to morphology

In chapter 3 we considered morphology, the structure of words. Clearly, if we are to make a sensible utterance we need to know how to put the words together. Knowing a language includes the ability to construct phrases and sentences out of morphemes and words.

Speakers of a language recognize the grammatical sentences of their language and know how the words in a grammatical sentence must be ordered and grouped.

All speakers are capable of producing and understanding an unlimited number of new sentences never before spoken or heard.

They also recognize ambiguities, know when different sentences mean the same thing and correctly perceive the grammatical relations in a sentence such as subject and direct object. Thus, the meaning of a sentence depends to a great extent on the structure or word order in which words occur in a sentence:

a. The hunter killed the lion.

b. The lion killed the hunter.

Sentence (1 a) does not mean the same as (1b), despite the fact that we using the same words in each case.

The part of the grammar that represents the speakers knowledge of these structures and their formation is called syntax. The term is derived from two Greek elements that equate to together and arrangement.

The aim of this chapter is to show what syntactic structure is and what the rules that determine syntactic structure are like. Most of the examples will be from the syntax of English, but the principles that account for syntactic structures are universal. Part of what we mean by structure is word order.

A principal distinction between morphology and syntax, then, is that the former is concerned with the internal composition of a word, whereas the latter is concerned with combinations of words. The distinction hinges on the word being a substantial concept. Nevertheless, some linguists (Poole 1999: 83) suggest that this may not be the case and if that view is accepted it undermines this distinction.

In any event there are problems. A function that is fulfilled by a single word in one case may be fulfilled elsewhere by a succession of words. Thus, where the English use two words, e.g. I give, some Romance languages do not usually use a subject pronoun (R. dau, It. do). Also, when referring to an action in the future, English speakers need an additional word to indicate future tense, whereas speakers of some Romance languages do not; the Italian equivalent of the English three-word phrase I will give is the single word darò

Even within the same language there may be elements that can be expressed either as words in a phrase or as morphemes in a word. There is little semantic or functional difference between more expensive and dearer.

If one regards the word as an inconsequential distinction it may be logical to dispense with it and to analyse sentences in terms of morphemes. One would thereby fuse the study of morphology and the study of syntax; many linguists do, indeed, consider morphology to be part of syntax.

4.2. The sentence patterns of language

The grammars of all languages include rules of syntax that reflect the speakers knowledge of these facts. The syntactic rules permit speakers to produce and understand an unlimited number of new sentences never produced or heard before, the creative aspect of language use.

The syntactic rules in a grammar must account for the following aspects:

4.2.1. The Grammaticality of Sentences: Grammatical or Ungrammatical

In English and in every language, every sentence is a sequence of words, but not every sequence of words is a sentence. Thus:

2. a. He put on the new coat.

b. He put the new coat on.

c. *He put the coat on new.

Sentences a. and b. are grammatical: this shows that sometimes a change of word order has no effect on meaning.

Even though the sequence in sentence c. is made up of meaningful words, it has no meaning.

Sequences of words that conform to the rules of syntax (i.e. to specific patterns determined by the syntactic rules of the language) are said to be well-formed or grammatical (sentences a. and b.). Those sequences of words that violate the syntactic rules are ill-formed or ungrammatical (sentence c. is ungrammatical because the sentence violates a rule of word order, namely the determiner must precede the noun).

4.2.1.1. What Grammaticality is Based On

Grammaticality judgements are not idiosyncratic or capricious but are determined by rules that are shared by the speakers of a language. Unconscious knowledge of the syntactic rules of grammar permits speakers to make grammaticality judgements.

The syntactic rules that account for the ability to make these grammaticality judgements include, in addition to rules of word order (as in 2.a. and 2.b), other constraints. For example:

3. a. The boy found the ball.

b. *The boy found in the house.

c. *The boy found quickly.

d. The boy found the ball in the house.

Speakers of English will place an asterisk in front of the second and third sentences. These two sentences are ungrammatical because the rules specify that found must be followed directly by a direct object (a noun like ball) but not by an adverb (like in the house or quickly)

Also consider the examples in (4):

(4) a. *Ann slept the baby.

b. Ann slept soundly.

Sentence (4a) is ungrammatical because the rules specify that the verb sleep occurs in a pattern different from find, in that it may be followed solely by an adverb (a word like soundly) but not by other kinds of phrases such as the baby.

Sentences are not random strings of words. To be a sentence, words must conform to specific patterns determined by the syntactic rules of the language.

4.2.1.2. What Grammaticality is Not Based On

The ability to make grammaticality judgements does not depend on having heard the sentences before. You may never have heard or read the sentences

(5) a. Enormous crickets in pink socks danced at the prom.

b. Colourless green ideas sleep furiously. (Chomsky)

but your syntactic knowledge tells you that they are grammatical.

Grammaticality judgements do not depend on whether the sentences are meaningful or not, as shown by the sentences above (5.a. and 5.b.).

For instance, sentence (5.b) doesnt seem to mean anything coherent, but it sounds like an English sentence, i.e. although the sentence does not make much sense, it is syntactically well-formed.

On the other hand, you may understand ungrammatical sequences even though you know they are not well-formed.

To most English speakers the sentence below (6)

6. *The boy quickly in the house the ball found.

is interpretable, or understood although those same speakers know that the word order is irregular.

On the other hand, grammatical sentences may be uninterpretable if they include nonsense strings, that is, words with no agreed-on meaning, as shown by the first two lines of Jabberwocky by Lewis Carroll:

Twas brillig, and the slithy toves

Did gyre and gimble in the wabe;

Such nonsense poetry is amusing because the sentences comply with syntactic rules and sound like good English.

Nor does grammaticality depend on the truth of sentences if it did, lying would be impossible. Untrue sentences can be grammatical, for example, sentences discussing unicorns can be grammatical, and sentences referring to pregnant fathers can be grammatical.

4.2.2. Ambiguity

Syntactic knowledge goes beyond being able to decide which strings are grammatical and which are not. It accounts for the double meaning or ambiguity, of expressions.

There are two types of ambiguities:

4.2.2.1. Structural ambiguities, i.e. ambiguities as a result of different structures. For instance, the phrase synthetic buffalo hides is ambiguous because it can have two meanings: it can mean buffalo hides that are synthetic or hides of synthetic buffalo.

This example illustrates that within a phrase, certain words are grouped together. Sentences have structure as well as word order. The words in the phrase synthetic buffalo hides can be grouped in two ways:

i. When we group: synthetic (buffalo hides) we get the meaning buffalo hides that are synthetic.

ii. When we group: (synthetic buffalo) hides we get the meaning hides of synthetic buffalo.

The two structures may also be illustrated by the following diagram:

synthetic buffalo hides synthetic buffalo hides

(i.) (ii)

The rules of syntax allow both these groupings, which is why the phrase is ambiguous.

Many sentences exhibit such ambiguities, often leading to humorous results. Consider the following sentence which appeared in classified ads (the example has been taken from V. Fromkin, 1993: 110)

1. For sale: an antique desk suitable for lady with thick legs and large

drawers.

In the above ad, the humorous reading comes from the two groupings:

i. (for lady with thick legs and large drawers) the legs and drawers belong to the lady, as opposed to the intended

ii. (for lady) (with thick legs and large drawers) - where the legs and drawers belong to the desk.

Because these ambiguities are a result of different structures, they are instances of structural ambiguity.

4.2.2.2. Lexical (or word meaning) ambiguities

Another ambiguity in sentence (1) is due to the two meanings of drawers: i. part of a piece of furniture; ii. womens underpants, knickers, panties.

Also consider the following sentence:

2. This will make you smart.

The two interpretations of this sentence are due to the two meanings of smart: clever or burning sensation.

4.3. Grammatical relations

4.3.1. Syntactic knowledge also enables us to determine the grammatical relations in a sentence, such as subject and direct object and how they are to be understood. Consider the following sentences:

1. a. Mary hired Bill.

b . Bill hired Mary.

c. Bill was hired by Mary.

In sentence (1.a.), Mary is the subject and is understood to be the employer who did the hiring. Bill is the direct object and is understood to be the employee.

In sentence (1.b.), Bill is the subject and Mary is the direct object, and as we would expect, the meaning changes so that we understand Bill to be Marys employer.

In (1.c.), the grammatical relationships are the same as in sentence (1.b.) but we understand it to have the same meaning as (a) despite the structural difference between (1.a.) and (1.c.).

Syntactic rules reveal the grammatical relations between the words of a sentence and tell us when structural differences result in meaning differences and when they do not.

We see that grammatical relations like subject and direct object do not always tell us who does what to whom since in (1a) and (1b) the grammatical subject is the who (the agent, performer), but in (1c) the grammatical subject is the whom (goal, recipient).

4.3.2. Theta/ -theory / Thematic roles

4.3.2.1. Theta theory is concerned with assigning thematic roles to the arguments of verbs. Theta is the name of the Greek letter θ, which corresponds to th in English, and since thematic begins with th it has become standard to abbreviate the expression thematic role to θ-role.

Its important to recognise that theme is being used differently here from its use in functional grammar, where it has largely a discourse meaning as the first item in a clause. In theta theory theme indicates one of a number of semantic roles which arguments fulfil. Clauses are seen as consisting of propositions, or logical statements, which require certain types of arguments in order to be acceptable sentences in English.

The approach is similar in some respects to M. Hallidays participant, process, circumstance model in functional grammar, and indeed, some of the terms overlap, but whereas Halliday is principally concerned with transitivity theta theory is more concerned with agency: who does what to whom.

The essential elements of the theory differ somewhat from linguist to linguist, but the following are the commonly assumed theta-roles:

(i) Theme (or patient) = entity undergoing the effect of some action

(The cat died)

(ii) Agent (or actor) = instigator of some action

(John threw the ball)

(iii) Experiencer = entity experiencing some psychological state

(John was happy)

(iv) Benefactive = entity benefiting from some action

(Mary bought some chocolate for John)

(v) Instrument = means by which something comes about

(John dug the garden with a spade)

(vi) Locative = place in which something is situated

(It rained in London)

(vii) Goal = entity towards which something moves

(Mary passed the plate to John)

(viii) Source = entity from which something moves

(John returned from London)

The value of incorporating thematic roles into a model of syntax is that it allows us to give a more principled account of the way in which linguistic items behave than relying simply on formal grammatical criteria. In the following pair of sentences, the phrase the vase fulfils the same grammatical role, that of subject, but two distinct thematic roles:

(i) The vase shattered the glass.

(ii) The vase shattered.

In (i) the vase is the cause of the shattering, hence it performs the role of instrument, whereas in (ii) it is the entity which undergoes the effect of shattering, hence it acts as the theme, or patient. The difference of thematic status is reflected in a difference of selection restrictions. In (i) we can replace the vase with the noise, or a hidden flaw, but not in (ii). Analyzing the thematic structure of these two sentences enables us to reveal differences which are not reflected in their constituent structure (see Immediate constituent analysis).

4.3.2.2. The meaning of a sentence is determined in part by the thematic roles of the noun phrases in relation to the verb.

These semantic relationships indicate who, to whom, with whom/what, from which, etc.

The noun phrase subject of a sentence and the constituents of a verb phrase are semantically related in various ways to the verb. The relations depend on the meaning of the particular verb.

For example, the NP the boy in The boy found a red brick is called the agent, or doer of the action of finding. The NP a red brick is the theme that undergoes the action. Part of the meaning of find is that its subject is an agent and that its direct object is a theme.

The NPs within a verb phrase whose head is the verb put have the relation of theme and goal. For example, in the sentence The boy put the red brick on the wall, the red brick is the theme and on the wall is the goal. The entire verb phrase is interpreted to mean that the theme of put changes its position to the goal. The subject of put is also an agent, so that in this sentence (The boy put the red brick on the wall) the boy performs the action.

The knowledge speakers have about find and put may be revealed in their lexical entries:

find, V NP (Agent, Theme)

put, V NP, PP (Agent, Theme, Goal)

The thematic roles are contained in parentheses. The first one states that the subject is an agent. The remaining thematic roles belong to the categories for which the verb is subcategorized. The direct object of both find and put will be a theme. The prepositional phrase for which put subcategorizes will be a goal.

The semantic relationships that we have called agent, theme and goal are among the thematic roles of the verb.

Other thematic roles are: location, source, instrument, experiencer, causative possessor, etc. (the list is not complete)

As we have already pointed out, the list of thematic roles differ somewhat from linguist to linguist. Thus, V. Fromkin and R. Rodman(1998: 176) give the following list of thematic roles:

Thematic role Description Example

Agent The one who performs the action Tom ran

Theme The one or thing that undergoes the action Tom called Ann

Goal The place to which an action is directed Tom put the book on the table

Location The place where an action takes place It rained in London

Source The place from which an action originates He flew from London to Paris

Instrument The means by which an action is He cut the bread with a knife

performed

Experiencer One who perceives something Tom heard Ann playing the

piano

Causative A natural force that causes a change The wind destroyed the roof

Possessor One who has something The tail of the dog got caught

in the door

Our knowledge of verbs include their syntactic category, how they are subcategorized and the thematic roles that their NP subject and object(s) have and this knowledge is explicitly represented in the lexicon.

Thematic roles are the same in sentences that are paraphrases. Thus, in both these sentences:

The dog bit the stick.

The stick was bitten by the dog.

the dog is the agent and the stick is the theme.

Thematic roles may remain the same in sentences that are not paraphrases, as in the following sentences:

The boy opened the door with the key.

The key opened the door.

The door opened

In all three of these sentences, the door is the theme, the thing that gets opened. In the first two sentences, the key, despite the different structural positions, retains the thematic role of instrument.

The three examples illustrate the fact that English allows many different thematic roles to be the subject of the sentence (that is, the first NP under the S). These sentences had as subjects an agent (the boy), an instrument (the key) and a theme (the door). The sentences below illustrate other kinds of subjects:

This hotel forbids dogs.

It seems that John has already left.

In the first example, this hotel has the thematic role of location. In the second, the subject it is semantically empty and lacks a thematic role entirely.

4.3.2.3. Thematic roles in other languages

If we contrast English with German, well see that German is much stingier about which thematic roles can be subjects. For example, in order to express the idea This hotel forbids dogs, a German speaker would have to say

In diesem Hotel sind Hunde verboten.

literally In this hotel are dogs (hounds) forbidden. That is, German does not permit the thematic role of location to occur as subject; it must be expressed as a prepositional phrase. If we translated the English sentence word for word into German, the result would be ungrammatical in German:

* Dieses Hotel verbietet Hunde.

Differences such as these between English and German show that learning a foreign language is not a matter of simple word-for-word translation. You must learn the grammar and that includes learning the syntax and semantics and how the two interact.

In many languages thematic roles are reflected in the (grammatical) case of the noun. The case of a noun refers to its morphological shape / form, often manifested as suffixes on the noun stem. English does not have an extensive case system, the only suffix being that of the possessive or genitive case (as in the boys book).

In highly inflected languages, such as Finnish, the noun assumes a morphological shape according to its thematic role in the sentence. For example, in Finnish, koulu is the root meaning school, and sta is a case ending that means directional source. Thus, koulusta means from the school. Similarly, kouluun (koulu + un) means to the school.

Some of the information carried by the grammatical case in languages like Finnish is borne by prepositions in English.

Thus, from and to often indicate the thematic roles of source and goal. Instrument is marked by with, location

Thus, koulusta means from the school. Similarly, kouluun (koulu + un) means to the school.

Some of the information carried by the grammatical case in languages like Finnish is borne by prepositions in English.

Thus, from and to often indicate the thematic roles of source and goal. Instrument is marked by with, location by prepositions such as on and in, possessor by of; agent, experiencer and causative by the preposition by in passive sentences.

The role of theme is generally unaccompanied by a preposition, as its most common syntactic function is direct object.

What has been called thematic roles in this section has sometimes been studied as case theory (Ch. Fillmore)

4.3.2.4. The THETA Criterion

A universal principle has been proposed, called the Theta criterion, which states in part that a particular thematic role may occur only once in a sentence. Thus, sentences like

*The boy opened the door with a key with a lock-pick

are semantically anomalous because two noun phrases bear the semantic role of instrument.

Also, in English the thematic role of possessor is indicated two ways syntactically, either as:

The boys red hat

or as: the red hat of the boy.

However, * The boys red hat of Bill

is semantically anomalous according to the Theta criterion because both the boy and Bill have the thematic role of possessor.

The syntactic rules permit speakers to produce and understand an unlimited number of sentences never produced or heard before, the creative aspect of language use. Thus, the syntactic rules in a grammar must at least account for:

i.            the grammaticality of sentences

ii.          word order

iii.        structural ambiguity

iv. grammatical relations

v. whether different structures have different meanings or the same meaning

vi. the creative aspect of language

A major goal of linguistics is to show clearly and explicitly how syntactic rules account for this knowledge.

4.4. Sentence Structure (Constituent Structure)

Syntactic rules determine the order of words in a structure and how the words are grouped. The words in the sentence

The child found the puppy

may be grouped into (The child) and (found the puppy) corresponding to the subject and predicate of the sentence.

A further division gives (the child) (found) (the puppy) and finally the individual words (the) (child) (found) (the) (puppy).

It is easier to see the parts and subparts of the sentence in a tree diagram:

The child found the puppy

the child found the puppy

the child found the puppy

the puppy

The tree is upside down with its root being the entire sentence: The child found the puppy and its leaves being the individual words the, child, found, the, puppy.

The tree conveys the same information as the nested parentheses, but more perspicuously.

The groupings and sub-groupings reflect the hierarchical structure of the tree.

The tree diagram shows among other things that the phrase found the puppy is naturally divided into two branches, the two groups, found and the puppy. A different division, say, found the and puppy, is unnatural in the sense that speakers of English would not use found the by itself or as an answer to the question What did you find? An answer might be the puppy but not found the. A word like found never occurs in a single group followed only by the.

Only one tree representation consistent with an English speakers syntactic knowledge can be drawn for the sentence The child found the puppy.

But the phrase synthetic buffalo hides has two such trees, one for each of its two meanings.

synthetic buffalo hides synthetic buffalo hides

synthetic buffalo hides synthetic buffalo hides

buffalo hides synthetic buffalo

(i.e. buffalo hides that are synthetic) (i.e. hides of synthetic buffalo)

Every sentence has one or more corresponding constituent structures composed of hierarchically arranged parts called constituents. These may be graphically depicted as tree structures. Each tree corresponds to one of the possible meanings. Structural ambiguity can be explicitly accounted for by multiple tree structures.

4.5. Syntactic categories (word classes)

Each of the groupings in the tree diagram The child found the puppy

is a member of a large family of similar expressions.

For example, the child belongs to a family that includes a police officer, your neighbour, this yellow cat, he, and countless others. Each member of this family can be substituted for the child without affecting the grammaticality of the sentence although the meanings of course would change.

The child found the puppy.

A police officer

Your neighbour

This yellow cat

He

A family of expressions that can substitute for one another without loss of grammaticality is called a syntactic category.

The child, a police officer and so on, belong to the syntactic category Noun Phrase (NP), one of several syntactic categories in English and every other language in the world. Noun Phrases may function as the subject or as various objects in a sentence, and only Noun Phrases may do so. NPs always contain some form of a noun (or noun substitute: pronoun).

There are other syntactic categories. The expression found the puppy is a Verb Phrase (VP). Verb Phrases always contain a Verb (V) which may be followed by other categories such as a Noun Phrase (NP) or Prepositional Phrase (PP). This shows that one syntactic category may contain other syntactic categories.

In addition to NPs and VPs, there are other syntactic categories such as: Sentence (S), Determiner (Det), Adjective (Adj), Noun (N), Pronoun (Pro), Preposition (P), Prepositional Phrase (PP) Adverb (Adv), Auxiliary Verb (Aux) and Verb (V).

n the sentence The child found the puppy: the child is a noun phrase, the head, the key word, being a noun. The noun phrase - the child constitutes a fundamental part of that sentence: the subject.

Some of these syntactic categories (word classes) have traditionally been called parts of speech.

All languages have such syntactic categories; in fact, categories such as Noun, Verb, Pronoun, and Noun Phrase are universally found in the grammars of all human languages.

The words child, puppy, neighbour, cat are classed as nouns. Nouns are words which denote something in the world around us, something inanimate, something animate like child, puppy, an attribute like strength, and so on.

Nouns are generally accompanied by a determiner, something which helps to identify what is being referred to. In the above sentences the nouns child and puppy are accompanied by the definite article the; other determiners are indefinite articles (a, an), possessive adjectives (my, your), numerals (three) and so on.

The word found, as used above, is a verb, a class that has such functions as indicating an action or a state of being.

Other classes are the pronoun, the adjective, the adverb, the preposition and the conjunction.

A pronoun replaces a noun and anything that may accompany it; thus a police officer or your neighbour could be substituted for he.

An adjective (e.g. yellow) provides more information about the thing or person indicated by the noun.

An adverb qualifies an adjective, a verb or another adverb.

A preposition is a class that indicates a relationship between other elements of the utterance.

A conjunction is a word which links elements of the utterance.

An utterance typically identifies something and then supplies some new information about that thing; the thing identified in our case is the child, the new information is found the puppy. The first element is the subject of the sentence; the second element is its predicate.

Similarly one can use the terms topic and comment and the terms theme and rheme.

A sequence with a subject and a predicate constitutes a clause. This may be a whole sentence, as in The child found the puppy. Alternatively, a sentence may consist of more than one clause, as in The parents are glad that the child found the puppy. In this latter example the second clause that the child found the puppy, is known as a dependent or subordinate clause.

A linguist analyzing the constituent structure of the sentence (The child found the puppy) will refer to the subject and a predicate as a noun phrase and a verb phrase. In linguistic notation: S > NP, VP.

This represents a deeper level of analysis than

S> art., N, V, art., N

Analysis at this deeper level allows us to define more general principles of sentence structure. Thus, we can say that a noun phrase precedes a verb phrase in a declarative sentence.

These two levels of analysis can be related to each other by constructing a tree diagram.

The elements on one level are constituents of those on a higher level, immediate constituents of these on the level immediately above. Words are immediate constituents of a phrase.

To understand the grammar of a language we need to know what can serve as constituents of an element. We need to know that a noun phrase must contain a noun or a pronoun. We need to know that a noun other than a proper noun needs a determiner which may be the definite article (the), an indefinite article (a or an) a demonstrative adjective (such as this), and so on. We need to know that a noun may be qualified by an adjective. We need to know that these all precede the noun.

Our awareness of what is and what is not a well-formed utterance helps us to understand almost immediately, and to discard as soon as possible false interpretations. As soon as we hear the definite article the we expect a noun, but not necessarily immediately. If the next word is very we expect an adjective to follow. If the word that follows that is old we are probably expecting a noun to complete the noun phrase. And so the process of decoding goes on.

A sequence with a subject and a predicate constitutes a clause. This may be a whole sentence, as in

The child found the puppy.

Alternatively, a sentence may consist of more than one clause, as in

The parents are glad that the child found the puppy.

In this latter example the second clause that the child found the puppy, is known as a dependent or subordinate clause.

Also, within the clause The child found the puppy we can insert another in order, for example, to further determine the child; we might say

The child who lives next door found the puppy.

Incorporating a clause within another clause in this way is embedding.

Lexemes in two or more word classes may share the same word form. Fear, for example, may be a verb or it may be a noun:

(a) The refugees fear persecution;

(b) He felt no fear.

To help us to identify which class a word belongs to we can enlist the aid of either morphology or syntax. Morphology shows us that in (a) fear is a verb; if the sentence were set in the past tense it would be this lexeme that acquired the past tense morpheme, realized by ed. Here syntax shows us very simply that fear is a verb; a sentence must have a verb and there is nothing else in this sentence that could fulfil that function.

Speakers know the syntactic categories of their language, even if they do not know the technical term.

In conclusion, syntactic categories are either phrasal categories such as NP and VP which can be decomposed into other syntactic categories or lexical categories, such as Noun and Verb which correspond to the words of the language.

Referring to word behaviour, G. Finch (2000: 134) points out that that there are two sets of criteria, namely, morphological and syntactic. Morphological criteria are concerned with the structure of words, with such processes as inflection. Most verbs will inflect to show tense (ask + ed), most nouns to indicate plurality (book + s), and many adjectives to show the comparative and superlative (hot>hotter>hottest). But not all. One of the difficulties of relying on morphology alone is that there is no one criterion which all words in a particular class will obey. As a consequence, linguists also use syntactic criteria, in particular, the distribution of a word in an individual string. Whereabouts a word can occur in a phrase is an important indication of its class.

4.6. Phrase structure

A phrase is a syntactic unit which typically consists of more than one word, and is intermediate between word and clause level in sentences. In most modern grammars it is regarded as the cornerstone of syntactic theory. In a phrase the individual words cohere together to form a single syntactic entity, capable of being moved around and also of being substituted by another word. In the following sentence, for example, the underlined words are capable of both of these:

The man went down the hill

Down the hill went the man (movement)

He went there (substitution)

The two tests which are being applied here are described by Nigel Fabb in the following way:

(i) if a sequence of words can be moved as a group, they may form a phrase

(the movement test); (ii) if a sequence of words can be replaced by a single

word, they may form a phrase (the replacement test). (Fabb, 1994: 3-4)

Phrases are formed out of the main lexical word classes. Indeed, we can see them as projections of these. So there are noun, verb, adjective, adverb, and prepositional phrases. The tests described above work better for some types of phrases than for others. Verb phrases, for example, are not very amenable to movement although they can be substituted for, albeit in a limited manner:

The man went down the hill and the dog did too.

In this example, did is substituting for the string went down the hill, and as a consequence is evidence for identifying it as a phrase.

Each phrase has a head word, taken from the word class which forms its basis. A noun phrase will have a noun as its head, a verb phrase a verb, and so on.

Because of the versatility and power of the phrase syntactically it is common nowadays to describe sentences in terms of the phrases which comprise them. Phrase structure grammar provides us with rules for sentence formation which utilize this form of analysis. The simple formula S NP + VP is a rewrite rule which captures the generalization that a basic sentence consists of a noun phrase plus a verb phrase.

The fact that The child found the puppy belongs to the syntactic category of Sentence, that the child and the puppy are Noun Phrases, that found the puppy is a Verb Phrase, and so on, can be illustrated in a tree diagram by supplying the name of the syntactic category of each word grouping.

These names are often referred to as syntactic labels.

The child found the puppy

S

the child found the puppy

NP VP

the child found the puppy

Det N V NP

the puppy

Det N

A tree diagram with syntactic category information provided is called a phrase structure tree. It is also sometimes referred to as a constituent structure tree.

This tree shows that a sentence is both a linear string (order) of words and a hierarchical structure with phrases nested in phrases.

Therefore, three aspects of speakers syntactic knowledge of sentence structure are disclosed in phrase structure trees:

The linear order of words in the sentence,

The groupings of words into particular syntactic categories,

The hierarchical structure of the syntactic categories: e.g., a sentence is composed of a Noun Phrase followed by a Verb Phrase, a Verb Phrase is composed of a Verb that may be followed by a Noun Phrase, and so on.

Every sentence of English and of every human language can be represented by a phrase structure tree that explicitly reveals these properties.

The phrase structure tree above is correct but it is redundant. The word child is repeated three times in the tree, puppy is repeated four times, and so on.

We can streamline the tree by writing the words only once at the bottom of the diagram. Only the syntactic categories to which the words belong need to remain at the higher levels.

S

NP VP

Det N V NP

the child found

Det N

the puppy

No information is lost in this simplified version. The syntactic category of each individual word appears immediately above that word. In this way, the is shown to be a Determiner, child a noun, and so on.

The lowest categories in the tree, those immediately above the words, are called lexical categories.

The larger syntactic categories, such as Verb Phrase (VP) are identified as consisting of all the syntactic categories and words below that point or node in the tree. The VP in the above phrase structure tree consists of syntactic category nodes V and NP, and the words found, the, and puppy.

The phrase structure tree also shows that some syntactic categories are composed of other syntactic categories. The sentence The child found the puppy consists of a Noun Phrase the child and a Verb Phrase found the puppy. The Verb Phrase consists of the verb found and a Noun Phrase - the puppy. The Determiner the and the Noun puppy together constitute a Noun Phrase, but individually neither is an NP.

4.7. Phrase structure rules

4.7.1. Grammars include phrase structure rules that specify the constituency of syntactic categories in the language.

The rules which govern the structure of utterances are known as phrase structure rules. Such rules allow for the generation of grammatical sentences in a language; they constitute a generative grammar for that language.

The sentences The child saw the teacher and The teacher saw the child are composed of the same words but they convey different messages. In the first of these two sentences we understand that it is the child that saw and the teacher who was seen, that it is the child that is the subject and the teacher that is the object. We understand this because English refers to the subject before it refers to the object.

Word order, then, is of extreme importance in indicating the relationship between the elements of a sentence. The customary word order may differ in different languages. Whereas English has the order subject verb - object (SVO). The Celtic languages (e.g. Welsh) put the verb first (VSO); Japanese puts the verb at the end (SOV).

Word order, then, is very important for the understanding of an utterance. This is particularly true in the case of a language such as English that is largely analytic. But there are other devices available for indicating syntactic function.

The German equivalent of the sentence The child gives the teacher the letter is Das Kind gibt dem Lehrer den Brief. The form of the definite article helps to indicate the relationship between the elements of the sentence. The das of das Kind indicates either a subject or a direct object, the den of den Brief indicates a direct object and the dem of dem Lehrer indicates an indirect object.

Where there is a relationship of this kind between form and function we may talk of a case system. Traditionally the subject form is referred to as the nominative case, the direct object form as the accusative case and the indirect object form as the dative case. In our German example only the articles reflect the case. In other languages with a case system there is usually greater variation in the form of the nouns; in Russian, for example, the equivalent of teacher is in the nominative case, in the accusative case and , in the dative case. Similarly, in Romanian the equivalent of teacher is profesorul in the nominative case, and profesorului in the dative case.

English has not always been as analytic as it now is; Old English had a well-developed case system and we have vestige of it in the possessive form of the noun. The Old English ancestor of father was fæder, this having a genitive form fæderes. The English have retained the case system, too, in the personal pronouns. While The child saw the teacher relies on word order for its meaning, The teacher saw the child meaning something quite different, we cannot change the meaning of He saw them by a simple rearrangement because he and them are marked for the functions of subject and object respectively.

As an alternative to The child gives the teacher the book we could say The child gives the book to the teacher. As an alternative to the teachers child we could say the child of the teacher. In each case we would be using a preposition to indicate the relationship between the elements concerned, the preposition to to indicate direction towards an indirect object and the preposition of to indicate possession. As English became more analytic, as it relinquished the case system, it had to rely more heavily on prepositions.

4.7.2. Grammatical trees must be accounted for in the grammar. Some kind of formal notation is required, preferably one that revels speakers knowledge precisely and concisely. To do this, linguists write grammars that include phrase structure rules that specify the constituency of syntactic categories in the language.

For example, in English a Noun Phrase (NP) can be a Determiner (Det) followed by a Noun (N). Thus, the NP sub-tree looks like

NP

Det N

the child

The phrase structure rule that makes this explicit can be stated as:

NP Det N

This rule conveys two facts:

A Noun Phrase can be a Determiner followed by a Noun.

A Determiner followed by a Noun is a Noun Phrase.

The left side of the arrow is the category whose components are defined on the right side. The right side of the arrow also shows the linear order of these components. Phrase structure rules make explicit speakers knowledge or the order of words and the grouping of words into syntactic categories.

The phrase structure tree of the previous section shows that the following phrase structure rules are part of the grammar of English.

VP V NP

This rule states that a Verb Phrase can be a Verb followed by a Noun Phrase.

4.7.3. Phrase Structure Rules in other languages

Other languages have different phrase structure rules, hence different tree structures. Thus, in Romanian, the Determiner may follow the Noun in some NPs, for example in copilul (child + the). The ungrammatical tree structure of English *N + Det (child + the) is a grammatical tree structure in Romanian.

In English phrase structure rule states that a Verb Phrase can be a Verb followed by a NP: found the puppy.

VP V NP

As can be seen from the rule, in English the verb is always the first member of the Verb Phrase. In other languages, however, for instance in German, the verb may occur in the final position of the Verb Phrase in some circumstances. German contains sentences such as:

Ich glaube dass Tristan Isolde liebt.

Which, if translated word for word, would be:

I believe that Tristan Isolde loves. (meaning I believe that Tristan loves

Isolde.

Despite these differences in detail, all grammars of all languages have the type of rule we are calling phrase structure rule, which characterizes the structure of phrases, sentences and syntactic categories of the language.

4.8. The lexicon

4.8.1. Speakers of any language know thousands of words. They know how to pronounce them in all contexts, they know their meaning and they know how to combine them in Phrases or Sentences, which means that they know their syntactic category. All this knowledge is combined in the component of the grammar called the lexicon.

Together with the phrase structure rules, the lexicon provides the information needed for complete, well-formed phrase structure trees.

The phrase structure rules account for the entire tree except for the words at the bottom. The words in the tree belong to the same syntactic categories that appear immediately above them.

Through lexical insertion words of the specified category are chosen from the lexicon and put into the tree. Only words that are specified as verbs in the lexicon are inserted under a Node labelled Verb, and so on. Words such as love, which belong to two (or more) categories, have separate entries in the lexicon, or are marked for both categories.

4.8.2. Subcategorization

The lexicon represents the knowledge speakers have about the vocabulary of their language including the syntactic category of words and what elements may co-occur together expressed as subcategorization restrictions.

The lexicon contains more syntactic information than merely the lexical category of each word. If it did not, speakers of English would be unable to make the following grammaticality distinctions:

a. The boy found the ball.

b. *The boy found quickly.

c. *The boy found in the house.

d. The boy found the ball in the house.

The verb find is a transitive verb. A transitive verb must be followed by a Noun Phrase, its direct object. This additional specification is called subcategorization and is also included in the lexical entry of each word.

Most words in the lexicon are subcategorized for certain contexts. Subcategorization accounts for the ungrammaticality of the following sentences:

a.      *Ann put the milk.

b.      * Jane slept the baby.

The verb put is a transitive verb which occurs with both a Noun Phrase (its direct object) and a Prepositional Phrase, as in:

a.      Ann put the milk on the table.

b.      Ann put the milk in the refrigerator

Sleep is an intransitive verb, so it cannot be followed by a Noun Phrase.

This information is included as the subcategorization of each word.

Other categories besides verbs are subcategorized.

For example, within the NP, if the Determiner is lacking (i.e. there is no article), only a plural noun (or a proper name) may be inserted (a.); and if the Determiner is a (the indefinite article), only a singular noun may be inserted (c.). This accounts for the following:

a.      Puppies love milk.

b.      *Puppy loves milk.

c.      A puppy loves milk.

d.      *A puppies love milk.

Subcategorization within the NP affects individual nouns. Belief is subcategorized for both a Prepositional Phrase (PP) and a Sentence (S), as shown by the following two examples:

a. The belief in freedom of speech

b. The belief that freedom of speech is a basic right.

Knowledge about subcategorization may be accounted for in the lexicon as follows:

A fragment of the lexicon Comments:

put, V, ------- NP, PP put is a transitive Verb which must be followed by both an

NP and a PP within the Verb Phrase

find, V, ------- NP find is a transitive Verb which must be followed by an

NP within the Verb Phrase

sleep, V, ------- sleep is an intransitive verb which must not be followed

by any category within the Verb Phrase

belief, N -------(PP), S belief is a Noun which may be followed by either a PP

or an S within the Noun Phrase

4.8.3. More lexical differences

Different words engender different syntactic behaviour, and this aspect of speaker knowledge is represented in the lexicon. The verbs of English occur in a wide variety of syntactic patterns. For example, the verbs want and force appear to be similar when we consider such sentences as

The conductor wanted the passengers to leave.

The conductor forced the passengers to leave.

but they differ in another syntactic context.

The conductor wanted to leave.

*The conductor forced to leave.

Try exhibits a third pattern differing from both want and force in that it is never followed directly by an NP:

* The conductor tried him to leave.

Try is, however, similar to want, but not force, in that it can be directly followed by an infinitive (i.e., the to form of the verb):

The conductor tried to leave.

These differing syntactic patterns of verbs are also specified in the lexicon, thus accounting explicitly for the knowledge speakers have about these verbs.

The examples given show only a single verb for each pattern, but each verb cited is representative of a class of verbs. For example expect, need, and wish pattern like want; allow, order, persuade pattern like force; and condescend, decide, and manage pattern like try.

Another instance of syntactically based lexical difference is found in the patterns in which believe and say appear, which are quite different from those of want-, force-, and try-class verbs:

The teachers believe Ann is outstanding.

The teachers say Ann is outstanding.

The teachers believe Ann to be outstanding.

*The teachers say Ann to be outstanding.

Both believe and say may be followed by a complete sentence. However, only believe can be followed by a sentence in which the verb occurs as an infinitive. As in the previous case, these patterns are representative of classes of verbs: suppose and think are like believe; forget and insist are like say.

The differences in syntactic patterns are part of the lexical representation of these verbs. The lexicon is a key component in the grammar, containing vast amounts of information on individual words.

4.9. Sentence relatedness

4.9.1. Sentences may be related in various ways:

i. For example, they may have the same phrase structure, differing only in their words, which accounts for their differing meaning.

For instance, the following two sentences:

The boat sailed up the river

2. A girl laughed at the monkey.

have the same phrase structure:

S

NP VP

Det N V PP

P NP

Det N

The boat sailed up the river

A girl laughed at the monkey.

ii. Two sentences with different meanings may contain the same words, in the same order, differing only in structure, like:

The boy saw the man with the telescope.

S

NP VP

Det N V NP PP

the boy saw Det N P NP

the man with Det N

the telescope

One meaning of this sentence the boy used a telescope to see the man is revealed by the phrase structure tree above. The key element is the position of the PP - with the telescope - directly under the VP, where it has an adverbial function and modifies the Verb saw.

In its other meaning, - the boy saw a man who had a telescope - the PP - with the telescope is positioned directly under the NP direct object, where it modifies the noun man.

S

NP VP

Det N V NP

the boy saw

Det N PP

the man

P NP

with

Det N

the telescope

These are cases of structural ambiguity. We can disambiguate the sentence if we move with the telescope in initial position.

With the telescope the boy saw the man.

Two different interpretations are possible because the rules of syntax permit different structuring of the same linear order of words, as revealed by the two phrase structure trees

iii. Two sentences may differ in structure, possibly with small differences in grammatical morphemes, but with no difference in meaning.

a. The girl wept silently. b. The girl silently wept.

a. Mary hired Bill. b. Bill was hired by Mary.

a. I know that you know. b. I know you know.

4. a. The astronomer saw a quasar b. With a telescope the astronomer

with a telescope. saw a quasar.

iv. Two sentences may have structural differences that correspond systematically to meaning differences.

5. a. The man is working. b. Is the man working?

6. a. The man can work. b. Can the man work?

7. a. The man will work. b. Will the man work?

8. a. The man is in the house. b. Is the man in the house?

The difference in the position of the verbal elements is, can, will corresponds to whether the sentence is declarative or interrogative.

Phrase structure rules account for much syntactic knowledge, but they do not account for the fact that Mary hired Bill has the same meaning as Bill was hired by Mary. Nor do they account for the systematic difference between statements and their corresponding interrogatives.

Since the grammar must account for all of a speakers syntactic knowledge, we must look beyond phrase structure rules.

4.9.2. Transformational rules

These correspondences are accounted for by transformational rules.

Consider the following sentences:

9. a. The girl wept silently. b. The girl silently wept.

as represented in these two phrase structure trees:

S S

NP VP NP VP

Det N V Adv Det N Adv N

The girl wept silently The girl silently wept

(a) (b)

Both sentences mean the same thing despite the position of the adverb. The phrase structure rules place the adverb at the end of the VP, thereby accounting for (a). How can (b) be accounted for? If we suggest a phrase structure rule with two optional adverbs, like this,

VP (Adv) V (NP) (PP) (Adv)

we are faced with the unacceptable possibility of generating ungrammatical strings such as:

*The girl cleverly wept bitterly.

A solution to this problem is to allow the phrase structure rules to generate the adverb in VP final position, and have another formal device, called a transformational rule, move the adverb in front of the verb, thus deriving the structure that represents (b) from the structure that represents (a).

The basic sentences of the language, whose phrase structure trees are called deep structures, are specified by the phrase structure rules. Variants on those basic sentence structures are derived via transformations.

The structures of sentences that we actually speak to which the rules of phonology are applied are called surface structures. If no transformations apply, then deep structure and surface structure are the same.

If transformations apply, then surface structure is the end result after all transformations have had their effect.

Thus, in sentence (1a) The girl wept silently the transformational rule moves the adverb in front of the verb: (1b) The girl silently wept.

Much syntactic knowledge not revealed by phrase structure rules is accounted for by transformations, which alter phrase structure trees by moving, adding or deleting elements.

In particular, families of structurally related sentences are revealed by virtue of having the same deep structure, with surface structure differences created by transformational rules.

A transformation, similar to the one that moves adverbs, moves prepositional phrases when they occur immediately under the VP. This transformation changes the deep structure of

The astronomer saw a quasar with the telescope

into the structure corresponding to

With a telescope the astronomer saw a quasar.

Transformational rules account for sentences whose surface structures are different but have the same meaning, such as sentence (2):

a. Mary hired Bill. b. Bill was hired by Mary.

Another transformation deletes that when it precedes a sentence in direct object position, but not in subject position, as illustrated by these pairs:

3. a. I know that you know. b. I know you know.

a. That you know bothers me. b. *You know bothers me.

The systematic relationship between statements (declarative sentences) and questions (interrogative sentences) in (4), (5) and (6) can be accounted for by a transformational rule. The rule refers to the movement of the Aux. (the auxiliary verbs: is, can, will) to the beginning of interrogative sentences.

If transformations apply, then surface structure is the end result after all transformations have had their effect.

In conclusion, to capture the knowledge speakers have about the syntax of their language, the grammar requires at a minimum, phrase structure rules, a lexicon richly endowed with speakers knowledge about individual words and a set of transformational rules describing the structure dependent patterning that occurs throughout the language.

4.10. Some approaches to syntax

4.10.1. Descriptive grammars

The most straightforward treatment of syntax is that provided by descriptive grammars. Descriptive grammars attempt to make precise, systematic statements about the syntax of a particular language. The basic methodology is simply to describe how the language works in practical terms. One of the most widely used of such grammars is R. Quirk et al.s, A Comprehensive Grammar of the English Language (1985).

The procedure adopted by descriptive grammars is quite different from that used by prescriptive grammars, which attempt to lay down rules about how people ought to speak and write rather than how they actually do.

Much of the terminology of the descriptive approach is to be found in traditional grammar, which is similarly concerned with analysing phrases and clauses. But modern descriptive approaches have taken account of recent developments in linguistics at a more theoretical level.

4.10.2. Immediate constituent analysis

The modern approach to syntax begins with the development of more explicit techniques of grammatical analysis, of which the most important was Immediate constituent analysis. This term was introduced by L. Bloomfield in his book Language. He illustrated the way in which it was possible to take a sentence (he chose Poor John ran away) and split it up into two immediate constituents (Poor John and ran away), these being in turn analyzable into further constituents (Poor and John, and ran and away).

In other words, a sentence is seen not as a sequence or a string of elements, Poor + John + ran + away, but as being made up of layers of constituents.

Constituent analysis is, therefore, the process of analyzing sentences into a series of constituents, which are organized in a hierarchical way. The major divisions made at a given level are called immediate constituents (or ICs); the smallest units resulting from this process of analysis are the ultimate constituents (or UCs). A grammar which analyses sentences in this way is called a constituency grammar.

Constituents are always represented hierarchically but the precise form in which they are shown varies among linguists. The most popular representation is in the form of a tree diagram.

But they may be represented in rectangular boxes as in:

the man ran through the park

the man ran through the park

man through the park

the park

park

Alternatively, parentheses can be used:

(((the) (man)) ((ran) ((through) ((the) (park))))

4.10.3. The post- Bloomfieldian school: Structuralist Approach

The techniques of IC analysis were developed and elaborated by some of L. Bloomfields followers, such as Zellig Harris.

Zellig Harriss Methods in Structural Linguistics, published in 1951, was the most rigorous attempt to develop formal techniques of description based upon the physical properties and distribution of units. George Trager and Henry Lee Smith Jr. published (also in 1951) An Outline of English Structure, in which the various levels of linguistic analysis were presented in a strict order phonetics, phonology, morphology, syntax- and it was argued that each level had to be studied independently of any other.

Both of these books exerted a great influence on linguistic thinking in the fifties, as did Charles Carpenter Friess The Structure of English, published in 1952, though it dealt solely with syntax. This book was the most systematic application of the structuralist approach to a language that had been attempted, and its importance lies both in its discussion of methodological and theoretical issues and in its detailed account of many aspects of English grammar in terms of the formal arrangement of items in structures.

We ought to add a reference to Archibald Hills Introduction to English Structures, published in 1958, which has the significant sub-title From Sound to Sentence in English, that reflects the structuralists methodology clearly.

H. A. Gleason Jr.s Introduction to Descriptive Linguistics, published in 1961, and Charles F. Hocketts Course in Modern Linguistics, published in 1958, both transcend the limitations of the structuralist account of language in many important ways, but were influenced strongly by it, and present very clear accounts of it.

It goes without saying that there were many differences between the various so-called structuralists: the post-Bloomfieldian school is by no means as homogeneous as the word school implies. But the point about treating meaning as secondary in linguistic descriptions is fundamental in much that they said. And the critical point which has to be made, and which has been made, frequently, by the generative grammarians, is that meaning is far too important and central to language to be left out or subordinated in this way.

Another, equally important thing that grammars have to do is take account of ambiguity in language. In IC analysis, however, disambiguation was impossible: an IC diagram either presented one meaning of a sentence only and ignored the others; or, it left us with a sentence analysis which was still ambiguous.

If we consider one of the examples which has now become a famous illustration of this point (it comes from Chomsky), the implications of this will be clear.

Chomsky takes the two sentences John is eager to please and John is easy to please, and points out that in traditional constituent terms, both sentences would get the same analysis. On the surface, they have the same structure. But underneath we all know if we think about it for a moment, that the word John has two different roles to play in the two sentences. In the first sentence he is the person who is doing the pleasing; In the second sentence, he is the person who is being pleased. Putting it in grammatical terms, John is the underlying subject of the first sentence, and the underlying object of the second.

4.10.4. Deep Syntax

4.10.4.1. Tagmemic theory: Tagmemics

Before Chomsky, there had been a number of critical reactions to aspects of the structuralist approach of the post-Bloomfieldian school.

The American linguist Kenneth Pike was responsible for a fresh theory of language which tried to get away from the over-concentration of the structuralists on classification and distribution at the expense of function. In his extensive book entitled Language in Relation to a Unified Theory of the Structure of Human Behaviour (967), Pike develops his tagmemic theory (tagmemics as it is usually called).

The tagmeme is a construct which tries to combine into one conceptual unit two ideas, which had previously been quite disassociated, the ideas of class and function. In Pikes terms, a sentence is analysed into a sequence of tagmemes, each of which simultaneously provides information about an items function in a larger structure, and about the class to which it belongs that could also fulfil that function. Such a structure is usually referred to as a series of slots into which various types of filler can go.

4.10.4.2. Scale and category grammar

A very similar theory, in some respects, is that associated primarily with the British linguist Michael Halliday. The theory is known as scale and category grammar. The reason for the labels scale and category is that they refer to the two basic ideas underlying the theory: the best way of accounting for languages structure was by postulating four major theoretical categories, and relating them via various scales of abstraction.

The categories comprised class (covering concepts such as verb and noun), unit (covering concepts such as sentence and clause), structure (covering concepts such as subject and predicate), and system (covering concepts such as the set of personal pronouns or tenses).

4.10.4.3. Systemic grammar

More recently, M. A. K. Halliday has developed a concept of systemic grammar, which would now be considered a quite distinct model. As its title suggests, it is based on the idea of system, which was a category of the early approach. This idea suggests that at any given place in a structure, the language allows for a choice among a small, fixed set of possibilities (we can have the/this/ my/ aman, for instance); and it is very similar to the Saussurean concept of paradigmatic relationships.

Language is viewed as a series of system networks, each network representing the choices associated with a given type of constituent (e.g. clause system network, nominal group system network, and so on).

In this approach, it is the clause system which is taken as the point of departure in analysis, not (as in most other models) the sentence

4.10.4.4. Stratificational grammar

Another theory is stratificational grammar developed by Sydney Lamb in his book entitled Outline of Stratificational grammar (published in 1966).

The theory is called stratificational because one of its chief features is its model of linguistic structure in terms of several structural layers, or strata.

Language for Lamb as for Halliday - is best viewed as a system of complex relationships which relates sounds (or, of course, their written counterparts) to meanings.

These relationships are not all of the same kind, however, but break down into more restricted sub-systems, each of which has its own structural organization; one sub-system operates at each stratum. There are three main strata hypothesized for all languages, semology, grammar and phonology; and these are subdivided in various ways to produce further systems.

4.10.4.5. Case grammar

Case grammar refers to an approach to grammatical analysis devised by the American linguist Charles Fillmore in the late 1960s, within the general orientation of generative grammar. The approach recognizes a set of syntactic functions (cases) in the analysis of a sentence, giving these an interpretation in terms of the semantic roles that these functions express, such as agentive, dative, and locative.

By focusing on syntactic functions, it was felt that several important kinds of semantic relationship could be represented, which it would otherwise be difficult or impossible to capture.

A set of sentences such as The key opened the door, The door was opened by / with the key, The door opened, The man opened the door with a key, etc. illustrate several stable semantic roles, despite the varying surface grammatical structures. In each case the key is instrumental case, the door is the entity affected by the action, and so on.

The term case is used because of the similarity with several of the traditional meanings covered by this term, but the deep structure cases recognized by the theory do not systematically correspond with anything in the surface morphology or syntax.

The original proposal set up six cases (agentive, instrumental, dative, factitive, objective and locative) and gave rules for their combination in defining the use of verbs. Later, other cases were suggested (Source, goal).

Case grammar exercised considerable influence on subsequent developments in linguistic theory.

4.10.5. Generative-Transformational grammar

Underpinning these theoretical perspectives has been a set of formal grammars, which set out not simply to describe, but to explain, how language works. Most influential here has been the work which as developed from the initiatives of the American linguist Noam Chomsky. Chomsky set out to establish a grammar that would make explicit the mental framework which speakers have in their heads for the production, or generation, of sentences.

Noam Chomskys ideas have dominated the development of linguistic thought since the 1950s. As professor of linguistics at the Massachusetts Institute of Technology (MIT) he developed the conception of a generative grammar, which departed radically from the structuralism and behaviourism of the previous decades.

His book Syntactic Structures, published in 1957, proved to be a turning point in 20^th century linguistics. Chomskys theory of language structure known as Transformational-generative (TG) grammar revolutionized work in linguistics.

Later, major publications on technical linguistic topics included Current Issues in Linguistic Theory (1964) and Aspects of the Theory of Syntax (1965). The latter publication introduced a new direction into generative theory and became the orthodoxy for several years.

The model of generative grammar Standard Theory, proposed by Chomsky in Aspects (1965) -, was viewed at the time as the leading statement concerning the aims and form of a transformational grammar

The model was revised in the early 1970s, when it came to be known as the Extended Standard Theory (EST) the extension being primarily in relation to the semantic rules, some of which were allowed to operate with surface structure as input.

A further revision in the mid-1970s, following developments in the notion of movement rules, was called the Revised Extended Standard Theory (REST).

A tenet of Chomskyan linguistics is the notion of competence. The term refers to a persons unconscious knowledge of the system of rules underlying his or her language. The notion of competence (contrasting with performance) has since led to the development of several related terms, notably pragmatic or communicative competence.

N. Chomsky recognized that surface structure is fairly arbitrary. The sentence John is eager to please and John is easy to please have a superficially similar structure but they convey a substantially different message; in the former John is doing the pleasing and in the latter he is on the receiving end.

For Chomsky it was essential to get below the surface realizations to the underlying principles, to the deep structure.

The generation of the surface forms was accounted for by transformational rules. Converting the active construction John wrote the letter into the passive construction the letter was written by John, conforms to the transformational rule

NP1 + V + NP2 > NP2 + aux + VPP + by + NP1

where aux refers to an auxiliary verb and VPP to the past participle of a verb.

Putting John writes a letter into the negative requires use of the dummy auxiliary verb do: John does not write a letter. This transformation might be described as

NP1 + V + NP2 > NP1 + aux + not +V +NP2

The proposed techniques have been adjusted over the years to meet concerns. The quest for the universal grammar and ultimately for the nature of mans innate disposition towards speech goes on.

4.10.5.1. Generative grammar

A generative grammar is one which sets out to specify, i.e. establish rules for, the formation of grammatical structures. Generative linguists are interested in how phrases, clauses and sentences are created. They aim to provide a rigorous and explicit framework that can produce (or generate) from a small number of general principles, or rules all the well-formed sentences of a language.

Generative grammar elaborates a set of rules, phrase structure rules, which encapsulate these formative processes. They are expressed in the form of rewrite rules, which are the basic blueprint for sentence formation.

Since Chomsky, generative grammar has grown, incorporating a number of different theoretical approaches and developments, which very considerably in their friendliness towards transformational methods of analysis. What they all have in common, however, is a rigorous empirical, and formalist approach to language.

Generative grammar has become increasingly more sophisticated over the years, with the extension into X-bar syntax, the aim of which is to simplify al the rules for phrase, clause, and sentence structure, into a few overarching rules using X as a category-variable, that is, a symbol which can stand for any word category. X-bar theorists argue that all languages have a version of X-bar syntax, and as such, it is an essential component of universal grammar.

4.10.5.2. Transformational grammar

Another important component of the Chomskyan version of generative grammar has been transformational grammar.

Noam Chomsky was interested in the way in which sentences with apparently identical surface forms could express completely different meanings. The classic case, frequently cited, is the pair:

John is eager to please

John is easy to please

A conventional tree diagram of these sentences would miss completely the difference between their logical and surface forms. The solution to this dilemma, suggested by Chomsky, was that underlying both sentences was a more basic level of structure, called deep structure. This was the primary grammatical level at which the sentences were formed, and represented their propositional core. The subsequent surface structures were seen as a consequence of a series of transformations which moved items into various syntactic slots.

The descriptive appeal of Chomskys approach is evident if we think of the way in which sentences are sometimes related:

(i) The man kicked the ball

(ii) Has the man kicked the ball?

(iii) The ball has been kicked by the man

(iv) Which ball has the man kicked?

There is clearly a semantic relatedness between these sentences. The person doing the kicking is always the same as is the thing being kicked. We could say then that all of these sentences derive from a common original, or deep structure, represented most nearly in the sentence The man kicked the ball. Linguists usually express the logical form of this in the following way:

kick (man, ball)

This representation means that there is a verb kick which has as its subject man, and its object, ball. To get from this abstract mental proposition to our starting sentence we have to imagine a minor transformation involving the addition of tense to kick and determiners to man and ball, plus some reordering. To derive the passives, negatives and interrogatives, above, however, would involve more complex transformations.

The consequence of Chomskys approach, then, is to suggest that we all possess two grammars as part of our linguistic competence. There is, first, a phrase structure grammar, which consists of the rules governing basic sentence formation, and secondly, a transformational one, which enables us to manipulate sentences to produce the full range of sentence types.

Transformational grammar has been enormously influential in recent years and has effectively changed the way in which most linguists approach syntax. But it has also been the subject of a great deal of debate and revision.

In the beginning, linguists tended to treat every sentence variation as a transformation with the result that anyone studying it encountered a plethora of complicated movement rules. It soon became clear, however, that there were so many exceptions to the rules that their power had to be constrained in some way.

An example of one such constraint is the rule for dative movement: this was supposed to relate pairs of sentences, like Tom gave a sandwich to Jane, and Tom gave Jane a sandwich. The transformation here was said to involve deleting to and moving the noun phrase following it Jane to a position immediately after the verb. The problem for this as a general transformation is that it is severely limited. There are some verbs, such as transmit, which will allow the first form but not the second:

They transmitted propaganda to the enemy

*They transmitted the enemy propaganda

And others, like ask, which will permit the second but not the first:

*John asked a question to Mary

John asked Mary a question

The constraints on this movement appear to be lexical rather than grammatical. As a consequence, any rule which we might formulate her would have to be verb-specific. Arguably, the place for this information is in the lexicon, rather than in a separate set of syntactic transformational rules.

Because of problems such as this, transformational grammar has, over the years, been streamlined to a few central operations. Not only that, but the terminology has changed. Deep and surface structure, have become D and S structure, principally because the original terms seemed to imply some sort of qualitative evaluation: deep suggested profound, while surface was too close to superficial.

Nevertheless, the principles of transformational grammar still remain very much alive in contemporary linguistics.

Chomsky himself has simplified the rules down to one general transformation: move alpha. In essence this means move anything. This may seem extraordinarily liberal but its operation is strictly controlled by a range of constraints which take into account a number of things, including the behavior of individual lexical items.

Some of these constraints take the form of additional theories which define the kinds of relationships possible within a grammar. Two of the most important of these are X-bar theory and theta theory. Others include government theory, binding theory, bounding theory, and control theory. Fundamentally, what has happened is that the responsibility for transformations has been distributed around the linguistic system rather than being located in one particular corner of it.

Inevitably, one consequence of this redistribution has been increasing pressure on transformational grammar to provide evidence for the remaining operations. Probably the most important of such evidence has to do with trace theory. According to this, constituents which are moved leave behind a trace, or echo of themselves in the surface structure.

To see how this works in practice consider the sentence:

Which ball as the man kicked?

Transformational grammarians would argue that this derives from the sentence The man has kicked the ball. First of all there has to be a rule which allows us to use which as an interrogative determiner replacing the. This may seem a bit odd to being with, but it isnt impossible to encounter it in ordinary conversation: The man has kicked which ball did you say? Then there are two major transformations. First the interrogative, which switches round the auxiliary verb has and the subject the man known as I (inflection) movement, and secondly, a wh transformation or wh movement that moves the nun phrase which ball to the front of the sentence, as in the diagram below:

The man has kicked which ball

I movement

wh movement

From early on, he argued that sentences had two forms, a deep structure (now D structure), which was the propositional core of the sentence, and a surface form/structure (now S structure), which was the way it appeared as an actual utterance. As such, two grammars were required. The first, a phrase structure kind, providing the underlying blueprint telling us, for example, that an abstract sentence comprises a NP (noun phrase) plus a VP (verb phrase) and the second, a transformational kind, specifying all the transformations needed to turn it into an actual sentence of English.

Some of these would be relatively minor, such as assigning number (singular/plural) to the noun, and tense (past/present) to the verb, and others would be major and result in the reordering of constituents (see immediate constituency analysis) to form interrogatives, passives, and negations.

Attitudes towards transformational grammar have been mixed over the years. Not all generative linguists are persuaded by the separation into deep and surface structure; Chomsky, though, has refined it considerably in recent times and it continues to be a potent influence in current work on syntax.

4.10.6. Functional / systemic grammar

A different kind of syntactic analysis is that provided by Functional grammar.

A functional grammar is one which seeks to derive syntactic structures from the functions which language is said to perform. All syntactic analyses take some account of functional categories. Terms such as subject and object, for example, are of this type. Functional grammar, however, attempts to discriminate, with a greater degree of delicacy, between different types of subjects and objects and relate these to semantic possibilities within the language.

In essence, the development of this kind of grammar was a reaction to the more abstract approaches associated with Chomskyan transformational grammar. The concern of functional linguists is with language as a system of meanings, rather than as an abstract code. They pursue the interactive, and communicative aspects of language.

Arguably the most influential of these linguists is the British linguist Michael Halliday, whose systemic functional model is widely used in stylistics and discourse analysis.

Halliday sees language as a sophisticated tool for accomplishing a number of central functions such as the need represent the world to ourselves and others, and the need to interact with other humans. What he attempts to do is to map these functions onto language. Instead of simply analyzing a sentence in terms of the phrases which comprise it, he is more concerned with the semantic role the phrases play in the communication of meaning. Sentences dont just contain subjects, and objects, but participants, each of which can be assigned a specific role, such as actor, or goal.

And verbs can be distinguished in terms of the processes they encode, whether action ones, such as running, and jumping, or mental ones, such as thinking and feeling. In this way he establishes a link between language as a code and language as a human tool.

Hallidays model of language is sometimes called systemic grammar. In simple terms this means that he sees language as a semantic system, i.e. a system for expressing meanings. At every point of the system a user is offered a series of choices which are both syntactic and semantic. The context within which these choices are made consists of three overarching functions which language is said to fulfil:

(i) the ideational function: the use we make of language to conceptualise the world.

This function emphasizes language as an instrument of thought, a symbolic

code, with which we represent the world to ourselves.

(ii) the interpersonal function: the use we make of language as a personal

medium. This function emphasizes language as an instrument of transaction by

which we represent ourselves to others.

(iii) the textual function: the use we make of language to form texts, whether spoken

or written. This function emphasizes language as an instrument of

communication with which we construct cohesive and coherent sequences.

According to Halliday, these functions relate to three central purposes which govern the form which clauses take. Clauses act as a representation (ideational function), an exchange (interpersonal function), and a message (textual function). Hallidays procedure is to take each of these in turn and describe the choices open to native users of the language.

I. The clause as representation:

This sees the clause as a means of representing the experiential world. As such,

it is composed of three functional components: participant, process and

circumstance. The participant function incorporates subjects and objects;

process incorporates the verb element; and circumstance incorporates

adjuncts /adverbials.

Centrally important in Hallidays model is the process component. It is this which

largely determines the types of participants which are possible. Halliday refers to

this as the system of transitivity and distinguishes six main processes: (i)

material processes; (ii) mental processes; (iii) relational processes; (iv)

behavioural processes; (v) verbal processes; (v0) existential processes.

II. The clause as exchange: this sees the clause as an interpersonal medium.

Principally important here is mood, that is the relationship speakers forge with

listeners through the form of the language. Traditionally, sentences are classified

as declarative, interrogative, imperative, and subjunctive. These forms

correspond to some of the speech acts which we use language to accomplish.

III. The clause as message: The message function of causes is connected very

much with their formation structure. Functional linguists characteristically

distinguish between theme and rheme, or alternatively, topic and comment.

The theme of a clause is its first major constituent, as in

John has done his homework.

The theme is the starting point of the clause and it s sometimes referred to as

the psychological subject

The rheme, or comment, is simply the remainder of the clause after the theme.

Its typical use is to expand on the theme and provide more information.

Theme and rheme overlap with another pair of terms, given and new.

4.10.7. Universal grammar

Theoretical grammars go beyond any individual language to try and establish what it is that all languages have in common. Most influential here is the concept of universal grammar. The assumption of grammarians working in this area is that linguistic knowledge is innate; in other words, that we are all wired in some way, or programmed to learn a language. The innateness hypothesis is well established now, and the search for what are termed language universals has yielded a number of insights into the way languages operate.

Universal grammar refers to the grammatical properties common to all languages and innately present in human beings. For many linguists, the discovery and description of these properties is the ultimate pinnacle of linguistic enquiry. Such properties as have been identified are called language universals.

There are two types of these: formal universals and substantive universals.

The formal kind of universal grammars have to do with the general design features of languages. We might say, for example, that all the operations which languages allow, involving the movement and ordering of items, are structure dependent and consequently, follow a structure dependency principle.

We must also implicitly be aware of another formal universal: that all languages are category-based. Every language has nouns, verbs, adjectives and adverbs, although the individual items which count as such may vary from one language to another. Formal universals, then, are very general principles which all languages obey, although they dont all obey them in the same sort of way (not every language forms questions in the manner English chooses to).

Substantive universals, on the other hand, are the particular structural features which languages have in common. In a sense, they are part of their linguistic content. All languages, for example, have structures such as noun phrase, and verb phrase, etc. and it is argued that they also have versions of X-bar theory and theta theory, which collectively determine how phrases combine into sentences.

As well as these universal principles, which serve to determine the broad grammatical structure of languages, there are also language-particular aspects of grammatical structure. Languages vary in the way they perform certain operations, but the interesting thing is that they do so in a principled manner. They vary according to certain parameters. One such parameter, for example, accounts for a distinction between those languages which allow sentences to be formed without a subject, and those which do not. In Italian, Parla Francese, is a well-formed sentence, but the equivalent in English, Speaks French, isnt. The difference is that Italian allows for an understood subject, he/she, whereas English doesnt. Languages like Italian, which have subjectless sentences of this type, are called pro-drop languages, and it has been found that they share a number of other characteristics, such as the ability to change the order of subject and verb in Italian, Falls the night is just as acceptable as The night falls.

Another parameter is the head (position) parameter. This has to do with the word-order variation within phrases.

The most recent development in Universal Grammar is a theory known as the principles-and-parameters theory (=PPT) of language. Given the research evidence which is mounting in its support, it looks set to become the dominant account of what a Universal Grammar might consist of.

Summary

Syntax is the study of the structure of utterances. It is commonly distinguished from morphology on the basis that syntax deals with combinations of words whereas morphology deals with the internal structure of words, but this distinction is not clear-cut.

To understand how words may be combined to construct utterances we need to distinguish between word classes.

Using word classes we can devise phrase structure rules.

We must recognize that a sentence is composed of groupings of words rather than just being a linear concatenation; by means of tree diagrams constituent structure analysis illustrates the groupings of words into phrases and clauses.

Noam Chomsky built on such an arrangement to ague that mankind has a universal grammar and to illustrate how that is transformed into a languages surface structures.

The means employed by a language to indicate the relationship between the elements of a sentence include word order, a case system and prepositions.

Sentences have structure that can be represented by phrase structure trees containing syntactic categories. Such a representation reveals the linear order of words, and the constituency of each syntactic category. Syntactic categories are either phrasal categories, such as NP and VP, which can be decomposed into other each syntactic categories or lexical categories, such as Noun and Verb, which correspond to the words of the language.

Phrase structure rules characterize the basic phrase structure trees of the language, the deep structures and include facts regarding syntactic constituency such as a Noun Phrase may be a Determiner followed by a Noun, but never(in English) a Noun followed by a Determiner.

The lexicon represents the knowledge speakers have about the vocabulary of their language, including the syntactic category of words and what elements may co-occur together, expressed as subcategorization restrictions.

Transformational rules account for sentences whose surface structures are different, but have the same meaning, such as Mary hired Bill and Bill was hired by Mary. They do this by deriving multiple surface structures from a single deep structure. Much of the meaning of a sentence is interpreted from its deep structure.

To capture the knowledge speakers have about the syntax of their language, the grammar requires, at a minimum, phrase structure rules, a lexicon richly endowed with the speakers knowledge about individual words, and a set of transformational rules describing the structure-dependent patterning that occurs throughout the language.

Exercises:

1. Do you think that morphology and syntax should be treated as separate areas of

study? Give your views and support then with reasons.

2. In the following sentence there is one example of the following word classes:

noun, pronoun, verb, adverb, preposition and possessive adjective:

Her dog always sleeps under it.

Rearrange the list of word classes to reflect the order in which they are

represented in the sentence.

3. Write five noun phrases in English. Use a different combination of word classes

in each case and incorporate a clause in one of your examples.

4. a) Construct a sentence in English that has the following structure:

S > art., adj., N, V, prep., art., N.

b) Represent the structure of your sentence in the form of a tree diagram.

5. Prepare a tree diagram for the following sentence:

The dog he bought likes young children.

6. Consider the following sentence:

The dog bit the man in the car.

The phrase in the car could be used

(a) to indicate where the biting took place or

(b) to specify that it was the man in the car that was bitten. How would the tree

diagram for each differ?

7. Explain the difference between the situations represented by the sentence in (a)

and the sentence in (b):

S S

NP VP NP VP

PP PP

N N V prep. N N V N prep. N

Copper shares sink after leak Copper shares sink after leak

(a) (b)

8. The very old fear

Write two sentences that begin with the above, one in which fear is a noun and

one in which it is a verb. Incorporate a prepositional phrase in each sentence.

9. *Us visit she on Sundays.

(a) Explain why the above is ungrammatical.

(b) Which personal pronoun in English would have the same form whether it

occupied the position of us or she in the above?

10. If English still had a case system, which phrase in the following sentence

would you expect to be (a) in the nominative case, (b) in the accusative case,

(c) in the dative case?

My youngest son threw the ball to the white dog.

11. Besides distinguishing grammatical from ungrammatical strings, the rules of

syntax account for other kinds of linguistic knowledge, such as

a. when a sentence is structurally ambiguous. (Cf. The boy saw the man with

a telescope.)

b. when two sentences of different structure mean the same thing. (Cf. The

boy wept silently and the boy silently wept.)

c. when two sentences of different structure and meaning are nonetheless

structurally related, like declarative sentences and their corresponding

interrogative form. (Cf. The boy can read and Can the boy read?)

In each case a. through c., draw on your own linguistic knowledge of English to

provide an example different from the ones in the chapter, and explain why your

example illustrates the point.

12. Paraphrase each of the following sentences in two different ways to show thatyou understand the ambiguity involved:

Example: Smoking grass can be nauseating.

i.            Putting grass in a pipe and smoking it can make you sick.

ii.          Fumes from smoldering grass can make you sick.

a.      Tom finally decided on the boat.

b.      The professors appointment was shocking.

c.      The design has big squares and circles.

d.      I cannot recommend him too highly.

e.      Terry loves his wife and so do I.

13. In terms of subcategorization, explain why the following are ungrammatical:

a. *The man located.

b.*Tom is fond that his children love animals.

c. * The children laughed the man.

Guide to exercises:

1. The distinction between morphology and syntax may presuppose that the

structure of a word is substantially different from the structure of longer

utterances. Such a distinction is undermined by the fact that semantically similar

utterances may consist of one word or more than one word.

2. Possessive adjective, noun, adverb, verb, preposition pronoun.

3. Examples might include a large selection, four very rich men, my young sons,

this book and the book that I prefer, the last of these incorporating a clause.

4. (a) The black dog sleeps in the garden.

S

(b) VP

NP

PP

NP

art adj N Vi prep art N

The black dog sleeps in the garden

5. S

NP

S VP

NP NP VP Vt NP

art N pron Vt Vt adj N

The dog he bought likes young children

6. In the case of (a) the prepositional phrase in the car would be an immediate

constituent of the verb phrase bit the man in the car. In the case of (b) it would

form part of the noun phrase the man in the car.

7. In the case of (a) there has been a fall in the value of copper shares while in the

case of (b) a policeman is having to wash in the same sink as somebody else.

8. Examples might include The very old fear of death persists to this day and The

very old fear the young in this city.

9. (a) Us is an object pronoun and she is a subject pronoun. This sentence

consequently contravenes the word order subject verb object. (b) You

10. (a) My youngest son (b) the ball (c) the white dog.