GSS

2.4 Sentence, clause and phrase

Traditional concept of sentence

In traditional grammars 'sentence' , like 'word' , is a basic though

largely undefined term. Sentences are thus simply 'composed of

words', and it is the function of syntax to state what words can

be combined with others to form sentences and in what order.

 How does Palmer see?

Most people are quite clear in their own mind that they know

exactly what a sentence is. This confidence arises because in a

literate society we are taught to indicate sentences in our writing

by putting in the punctuation. The normal mark of the sentence

is the full stop; it would be an error of punctuation to mark the

end of a sentence with a comma. In addition new sentences must

begin with capital letters. It certainly does not give us a

definition.

 In fact, we are taught at school to recognize sentences

through practice, not by a set of rules.

The traditional grammars, however, sometimes provide a kind

of definition: a sentence is the 'expression of a complete thought' .

But this shares all the faults of the notional

definitions that we discussed in Chapter 1. How do we know

what a complete thought is? Is 'cabbage' or 'man' a complete

thought? If not, why not? And is If it rains, I shan't come one

thought, or two joined together? It would seem quite impossible

to provide any definition along these lines.

One such would be that it contains a subject and predicate - that on the one hand it indicates something that we are talking about(subject),

 And on the other it says something about it(predicate)

For instance, in John iscoming we are talking about John, the subject,

and also saying that he is coming, the predicate.

 The difficulty  here is that, if this definition is to be of any use, we must be able to identify what we are talking about, and very often we talk about several things at once.

What is Subject then?

For instance, in the sentence John gave the book to

Mary we are clearly talking about John, the book and Mary and

all three might seem to be the 'subject' in this sense.

A natural reaction, especially from someone who has learnt some traditional

grammar, might be to say that we are talking about John

and that what we are saying about him is that he gave the book

to Mary. But this begs the question.

 It defines the subject as the

grammatical subject, and the grammatical subject can only be

defined in terms of the sentence. Moreover the grammatical

subject often does not indicate what we are 'talking about'. In

The birds have eaten all the fruit it is probable that what we are

talking about is the missing fruit and not the unidentified birds!

More strikingly, in It's raining what is the subject? It? But what

is 'It'? - the weather, the universe, or what?

As we saw in

Chapter 1, although sections of speech are often marked by

intonation, it is not the case that every intonation tune will mark

a stretch of speech that, in written form, would begin with a capital

letter and end with a full stop.

Palmer’s objection

Moreover a great deal of spoken

language does not consist of sentences in the sense in which the

term is understood for writing at all. Much of it is made up of

incomplete, interrupted, unfinished, or even quite chaotic sentences.

Speech may be made up of utterances, but utterances

seldom correspond to sentences. We could not, for instance,

identify all the sentences in a conversation that went:

MARY: John! Coming?

JOHN: Yes dear, fwas only-

MARY: Oh do hurry up and - we ought to catch the bus - only they don'talways run on time - wretched people - as long as you're quick. I’ve been ready for some - since half past seven.

Such a conversation is not abnormal; much of our everyday

speech is like this.

Palmer’s suggestion

A linguistic definition of the sentence must be in terms of its

internal structure. A sentence will be composed of certain

specified elements in a certain order, ultimately, of course, of

words or parts of words. A statement then of the structures will

provide us with a definition of the sentence.

 For instance, we

might argue that the basic sentence structures of English are of

the type NV, NVA, NVN, NVNN. Examples would be John

came, John is good, John saw Bill, John made Bill president.

came, John is good, John saw Bill, John made Bill president.

(This is by no means a complete list.)

All other sentences could

be regarded as derived from these by either addition, e.g. of

adverbs - John came quickly - or by expansion (see pp.

122-4): instead of John we could have the boy, the little boy, the

silly little boy and even the silly little boy on the other side of the

room.

But this is not really satisfactory. We can say that a

sentence is a linguistic item that has the structure we assign to

the sentence.

 But why these structures? In particular, why not

much larger structures? Why do we not want to consider It's

raining, I'm not going out as a single sentence? The answer is

that these sentence structures are the largest that can be handled

in a grammatical description. We can make an accurate statement

about the limitations on the co-occurrence of the items in the sentence, but we cannot with any accuracy deal with larger

structures.

What does Bloomfield say?

This was put quite clearly by Bloomfield, who defined

a sentence as 'an independent linguistic form, not included by

virtue of any grammatical construction in any larger linguistic

form'. He considered the example: How are you? It's a fine day.

Are you going to play tennis this afternoon? and went on to show

that there are no grammatical restraints linking these into a

single structure; they have to be regarded as three separate

units, that is to say, three sentences.

Palmer’s stand on Bloomfield’s model

The sentence is, then, the largest unit to which we can assign

a grammatical structure. Nevertheless, it would be an error to

believe that outside the sentence there are no restraints, no

features that link one sentence to another. There are, on the

contrary, plenty. Many words such as however, therefore, later,

other serve very often to refer from one sentence to another.

More striking perhaps are what are sometimes called the 'proforms'

of a language. Pronouns are familiar enough. He, she and

it may 'stand for' the man, the woman, the table, etc.

We find in one sentence The man . .. but in the next, He ... But there are also 'pro-verbs'.

Did in John came and so did Mary. did

stands for came - Mary came.

All the auxiliary verbs in Englishcan act as pro-verbs in the sense that they alone stand for the whole of the verbal element of which they are or were only the

first word:

John is coming.

I haven't seen him.

Must you come?

He'll have been there.

Is he?

But I have.

I really must.

No he won't.

The verb DO is particularly important because it is the pro-form

used where there is no auxiliary verb:

He came yesterday. No, he didn't.

Palmer and Paragraph

Beyond the sentence there is no clear limit at all. In writing

we use paragraphs, but what are the rules for paragraphing?

There are, perhaps, some vague rules - that we start a new

paragraph where we start on a new subject - but one may

well suspect that paragraphs are also dictated by purely

aesthetic considerations;

 pages without paragraphs look uninteresting.

We· may not, perhaps, like Alice, demand conversation

or pictures in our books but we like paragraphs.

Incomplete, interrupted 'sentences'

A problem is raised by the incomplete, interrupted 'sentences'

that we discussed earlier (p. 67). Some linguists have argued

that they should be analysed independently and treated as

possible structures of the English language. But this would seem

to be a mistake, above all because there would then be an

infinite number of structures and no grammar could claim even

partial completeness. There are, however, three kinds of' incomplete

sentence' .

First, there are those that are caused by interruptions or

changes of mind on the part of the speaker. In the imaginary

conversation on page 61 we find examples in I was only - and

I've been ready for some - . These raise no problems for

grammar; they are genuinely incomplete sentences, understandable

and analysable as such. (The linguist may not be altogether

uninterested in them, however; he may well want to know

whether there are conditions for interrupting, for hesitation,

change of mind, etc.)

Secondly, there are incomplete sentences that are dependent

on what has gone before. John, for instance, might be a reply

to Who did it? or Who did you see? It can therefore be

reasonably understood as an incomplete form of John did it

or I saw John. These make extensive use of pro-forms and are

to be analysed in terms of the complete, ~xpanded, 'original'

form. They are 'contextually' conditioned and can only be

understood as such.

Thirdly, there are incomplete sentences such as Coming?

Coming! Found them? Got you! which might seem equally to be

shortened forms of Are you coming? I'm coming! Have you

found them? I've got you! But these are not contextually

conditioned; they do not in any way depend on what has gone

before. There is a case for treating them as English sentences in

their own right, but there is also an argument for treating them

as derived from the longer forms by 'deletion' - we 'delete' the

pronoun and the auxiliary verb.

What about Phrase?

The sentence consists of words, but the words are grouped

into elements that are smaller than the sentence. For these most

linguists use the term 'phrase'. Sentences are thus analysable

into phrases. The most important phrases of the sentences are

the verb phrases and the noun phrases (symbolized as VP and

NP respectively), e.g.

John likes Mary (NP VP NP).

(However, the term VP is used in a rather different sense in the

phrase structure analysis of transformational generative grammar

- see 4.2.) A phrase in this sense can be a single word, but

the phrases are often much longer than single words:

The little boy has been reading a fairy

story (NP VP NP).

In addition there are elements within the sentence such as this

morning or in the garden which are sometimes called 'adjuncts'

but are better called 'adverbial' phrases.

If we consider the simplest

phrases (but see below, pp. 142-3) of English, we find that a

noun phrase consists either of a pronoun alone (or, rarely, with

an adjective, e.g. Poor you!), or of a noun preceded by various

words some of which are adjectives and other determinatives

(the, this, my, etc.), and sometimes followed by a word such as

abroad or asleep (people abroad, children asleep). In fact the

modifiers of the noun phrase, all the words that is to say except

the noun itself, are of numerous and varied types. In particular

they have their own place in the sequence. Not only can we not

place asleep before the noun (*asleep children), but we have to

put the adjectives in the right order (little red hen, not *red little

hen) putting also any other elements before or after the

adjectives and in their right order. This is clearly shown by the

following sequences which permit little or no variation:

All the twenty-five little English children.

Both her worn-out red cotton dresses.

The study of the noun phrase itself is worthy of a complete book.

So too is the verb phrase. Its structure is a little less complex

in some ways, somewhat more complex in others. The maximum

length of a verb phrase seems to be five words, e.-g.:

He may have been being beaten,

"S'-

though it may be doubted whether all five often occur together.

There are certainly five elements that occur in sequence:

(1) a modal followed by the simple form of a verb.

(2) HAVE followed by the past participle (the perfect).

(3) BE followed by the -ing form of the verb (the 'progressive'

or 'continuous').

(4) BE followed by the past participle (the passive).

(5) the main verb.

It is in this context, the analysis of sentences into NPs and

VPs, that we can talk of 'subjects' and 'objects'. Instead of

treating John likes Mary as NP-VP-NP, we can describe it as

subject-verb-object, and say that John is the subject and

Mary the object of the verb likes. It is, however, a little

misleading to use the term 'verb' at this level, and so to

analyse John likes Mary as subject-verb-object. For the term

'verb' is used as the name of the word class, like 'noun'.

Strictly, we need another term for the sentenc~ element, and

PREDICATOR has been suggested. The sentence can either be

treated as N-V-N (or, more strictly, as NP-VP-NP) or as

subject-predicator-object. Unfortunately, the term 'verb' is regularly

used in this other sense, and this practice will be followed

with reluctance; but it should be remembered that 'verb' is

ambiguous, referring either to the word class or the sentence

element. We shall return to subjects and objects later (pp. 75-7).

One further traditional distinction IS between TRANSITIVE and

INTRANSITIVE sentences, transitive sentences being those with

objects (John likes Mary) and intransitive those without (John

sings) (and we may similarly refer to the transitive and intransitive

verbs LOVE and SING). This distinction is related to voice

(p. 88), since only transitive verbs in English may have passives.

We may also, perhaps, talk of di-transitive verbs, those that have

two objects (John gave Marya present); here Mary is described

as the INDIRECT object and a present as the DIRECT object. In

some languages even further distinctions can be made (p. 76).

But there are, in English and most other languages, many other

sentence types for which no similar names are available (see p.

74).

Clause

Traditional grammars also talk of 'clauses', which are 'sentences

that are part of larger sentences'. This definition is,

strictly, self-contradictory, but it still indicates what is meant,

and illustrates a very important characteristic of natural

languages. In, for instance, John stood still and Mary ran away

we have a 'larger-sentence' consisting of two sentences joined

together by and. Similarly in While John was standing there Mary

ran away there are two sentences, the first introduced by while,

making up the larger sentence.

However, these two larger sentences illustrate two different

ways in which sentences may be joined together. In the first,

they are simply linked by and, and we can link as many as we

wish in this way. Moreover, the relationship between the two

sentences is not very different from that of two sentences

separated by a full stop. There would be little difference in, for

instance, John stood still. Mary ran away. (It is not true, however,

that we can link any two sentences with and. We cannot say

* Come here and John has arrived though we can say Come here.

John has arrived. But, for the most part, there are few restrictions

on sentences joined by and.) This kind of linking of sentences is

known as 'coordination' .

The second way in which two sentences may form a larger

sentence is one in which, instead of the two sentences being

joined together as equals, one of the sentences functions as part

of the other. For instance, alongside He said many things, we

can say He said that he was coming. Clearly that he was coming

has the same kind of function as many things, and is, perhaps,

the object of He said ... A term used for this today is

'embedding', one sentence being embedded within another.

 The traditional grammars referred to this as 'subordination' and

talked about the embedded sentence as a 'subordinate clause'.

These subordinate clauses were further classified into nounclauses,

adjective-clauses and adverb-clauses, according to

whether they had the function of nouns, adjectives or adverbs

within the other sentence (the 'main' clause). For instance, in

the example we have just mentioned that he was coming has the

function of a noun, for it is nouns and noun phrases that act as

objects. An example of an adjective clause would be who was

standing there in The boy who was standing there ran away. It has

a function similar to that of little in the little boy, though the rules

of English permit little to come before boy but the adjective

clause to come after it. An a,dverb clause would be while I was

standing there, which has the':same kind of function as yesterday

in I saw John while I was standing there.

The traditional grammars ritserve the term sentence for the

larger or 'maximal' sentence and talk about the sentences of

which it is composed, the 'minimal' sentences, as clauses. This js

an important distinction, since there are features of the clause

that are not features of the larger sentence. Reflexive pronouns,

for instance (p. 59), will normally refer to a noun within the

same clause (but not to one in another clause in the same

sentence), as shown by

The boy said that John had hurt himself.

Here himself can refer only to John, not to the boy.

The grammars make a distinction, moreover, between

'clause' and 'phrase', though not using 'phrase' in the sense in

which I have used it (to distinguish the essential parts of a

sentence - the noun phrase, the verb phrase, etc.), but to

refer (amongst other things) to a special kind of embedded

sentence - one without a finite verb. A 'finite verb' is a verb

form that can stand alone in an independent sentence - comes

is finite but coming is not, since we can say He comes, every

day but not * He coming every day. We are told therefore that

how to do this in I don't' know how to do this is a noun

phrase, not a noun clause, because it has no finite verb. But

thIs seems an unimportant distinction.

 'There are all sorts ofrules for embedding or subordination,

but what is importantis that the embedded sentence has still many

 of the characteristics of a sentence.

In the example above we still have a

predicator do and object this. In I don't like John doing that

we have John (subject) doing (predicator) and that (object)

- an almost normal sentence, but without a finite form of a

verb. We shall discuss some of these problems again later.

In this chapter we have talked about 'analysing' sentences.

Traditional grammar made analysis or 'parsing', as it was often

called, an essential exercise. In Nesfield, for instance, we are

instructed to divide a sentence first into subject and predicate,

then to divide the subject into nominative and its enlargement

and finally its predicate into finite verb, completion and extension,

the completion being either object or complement or both.

For the sentences The new master soon put the class into good

order and A bird in the hand is worth two in the bush the analysis

is;

This indicates in some degree the structure of the sentence,

but is, even within its own lights, far from satisfactory. Why, for

instance, do we have enlargement only for the nominative? The

enlargement includes all of what today would be called the

modifier of the noun - the article and the adjective, etc. But all

nouns in the sentence may have similar modifiers too. The occurs

as a modifier in the class which is the object, and in the bush

which is part of the complement, and nouns can equally occur in

the extension as part of prepositional phrases. It is misleading

too to talk about 'completions' which are required, according to

Nesfield, because some verbs 'do not make sense in themselves'

but need either objects (the- transitive verbs) or complements

(the copulative verbs). For there are verbs which seem also to

require extensions. An example is to lie (as in to lie down) which

needs such extensions as there or on the table (it lay there/on the

table). In fact, if we investigate carefully we shall find that verbs

can be classified into a number of different types, each requiring

a different set of following grammatical elements. We need to

recognize at least seven sentence structures, exemplified by;

(1) The man smiled. (NV)

(2) John seems happy. (NVA)

(3) The woman hit the man. (NVN)

(4) I gave the boy a book. (NVNN)

(5) The boy sat on the floor. (NVPN)

(6) The girl made John hapPY. (NVAN)

(7) John put the book on the table. (NVNPN)

Also, it would usually be said that, in spite of their superficial

similarity to (3) and (4), different analyses are required for:

(8) The woman became a teacher.

(9) 1 made the boy their leader.

For (8), part of the argument would be that there is no passive,

as there is for (3) (The man was hit by the woman but not * A

teacher was become by the woman), and for (9) that we cannot

say *1 made their leader to the boy, by analogy with 1 gave a book

to the boy. Such arguments take us further than the immediately

observable structures.

Let us now return briefly to subjects and objects. It is clear that

these are not merely elements or constituents of sentences, as N Ps

and VPs are, but that they are essentially functional or relational

(and they have been described as 'grammatical relations').

We can and must define subjects and objects formally. In

English the relevant criteria are position in the sentence,

agreement of the subject with the verb and morphology in the

case of pronouns (1 vs. me, etc.). There is some temptation also

to define the subject as the 'actor', the person who performs the

action, and the object the' goal' or 'recipient' , the person or thing

that is affected by it. But this wo1,1ld not allow us to identify the

subject as John in any of the following sentences, for in none of

them is John 'acting' in any intelligible sense:

John suffered terribly.

John looked sad.

John saw his brother.

John sank under the waves.

It would be equally impossible to determine the subject in

John lent a book to Bill.

in view of

Bill borrowed a book from John.

Who is the actor, who the recipient? If John is the subject in the

first sentence, Bill cannot be the subject in the second as long as

we rely on purely notional definitions. But there is no real doubt

in linguistic terms - in terms of position in the sentence.

Nevertheless, although we cannot define subjects in terms of

being agents, we can say that subjects are typically agents, just

as plural typically refers to more than one (p. 35). That is to

say, where we can establish subjects on formal grammatical

grounds, we shall find considerable, but not absolute, correlation

with the notion of agency.

Naturally, not all languages mark subjects and objects in the

same way. In Latin, for instance (see the example on p. 20),

word order is not a defining characteristic, while case-marking

is. In other languages, e.g. Swahili, the main criterion is agreement

of both subject and object with the verb. Nor are subject

and object the only grammatical relations. In English we need

to distinguish direct objects and indirect objects (see p. 71),

while in other languag~s instruments and beneficiaries are clearly

marked formally.

More surprisingly, perhaps, it is by no means certain that the

relations of subject and object are appropriate for the description

of all languages. There are probably few languages, if any, that

do not have some means of indicating actors and goals. Yet it

has been reported that there is at least one (Lisu, Lolo-Burmese)

that does not mark this distinction in any regular fQImal way, so

that a single sentence may mean either 'People bite dogs' or

'Dogs bite people'. This seldom creates any ambiguity, however,

since either the ·context or common sense will usually point to

the correct meaning; where ambiguity is likely, it is always

possible to add some extra comment.

More important are the languages with ERGATIVE systems.

These differ from our familiar subject-object system in a

fundamental way. The need to distinguish subjects and objects

arises only where there are two NPs in a sentence; there is no

need for the distinction where there is only one NP - in an

intransitive sentence. Yet English and other familiar languages

always mark this single NP in the same way as the subject of the

transitive sentence. In an ergative system the single NP is

identified with the other NP of the transitive sentence - the one

we should call the 'object'. It is rather as if English said. * Him

sings instead of He sings. An example from Dyirbal (Australia)

is:

IJ uma banaganyu.

yabu banaganyu.

IJuma yabU1}gu buran.

yabu IJuma1}gu buran.

'Father returned' .

'Mother returned'.

'Mother saw father' .

'Father saw mother' .

Notice that in the third and fourth sentences it is the words for

'father' and 'mother' respectively that have no endings (as in the

first and second), while the ending -T)gu appears on 'mother' and

'father'. If we continue to use the terms 'subject' and 'object'

here there are grave problems. We could say that intransitive

sentences have only objects and no subjects. Alternatively, we

might simply say that the subjects of intransitive sentences are

marked in the same way as the objects of transitive ones. But

this would be paradoxical and difficult to explain; it would be to

impose our own familiar system upon an unfamiliar one, just as

Latin grammar was imposed upon all kinds of languages. Instead,

it is usual now to refer to the NP of the intransitive sentence

together with the 'object' of the transitive as ABSOLUTIVE, and

to the other NP of the transitive, its 'subject' ,as ERGATIVE.

Reference was· made earlier to 'ergative systems' , not, as is usual,

to 'ergative languages', because, curiously, the languages never

seem to use the ergative system throughout all their syntax.

Nevertheless, languages with ergative systems are found all over

the world and should neither be ignored nor forced into the

subject-object mould. They include Basque, Eskimo, Caucasian

languages such as Georgian, North Indian languages such as

Hindi and Bengali, and many languages in Australia.

TENSE we move to consider a category associated directly with the verb. We have seen already that tense is often, though very misleadingly, associated with time, and the problem of

tense in English is dealt with in some detail in Appendix C.

There are few, if any, languages in which there is a category that

is totally related to time, though there are some, e.g. Latin,

where a division into present, past and future is justified on

formal grounds with some fairly clear relationship to time, e.g.

amo 'I love', amabo 'I shall love' , amabam 'I loved'; but even

here amabam is perhaps better translated as 'I was loving' with

amavi as the 'I loved' form.

In many languages there is ASPECT as well as, or instead of, tense, tense supposedly referring to time and aspect to completion, duration and similar concepts. In Latin, we can treat the distinction between the perfect tenses and the others as one of aspect. In the Slavonic languages a regular distinction is made

between verbs referring to completed and those denoting noncompleted action; Russian has to distinguish between reading a book but Dot finishing it (citat') and reading a book and finishing it (procitat'). In Classical Arabic the only distinction in the verb seems to be one of aspect, complete and incomplete.

Morphologically, English has only two tenses, past and

present, as illustrated by takers) and took or lovers) and loved.

The traditional 'future' tense is formed with the auxiliary verbs

WILL and SHALL (see below and Appendix C). But English also

has progressive (or 'continuous') forms expressed by the auxiliary

verb BE followed by an -ing form as in:

The boy is reading a book.

It also has perfect forms which are expressed by the auxiliary

HAVE and a following past participle:

The boy has read the book.

The term 'aspect' is often used to refer both to the progressive

and to the perfect, though the term 'phase' has been suggested

for the latter. In any case, it is important to realize that these are

only labels. It is unwise to attempt to find clear semantic

distinctions between aspect and tense (and phase), except where

a language, such as English, has more than one such category.

In other languages it is by no means clear always whether 'aspect' or 'tense' is the more appropriate label for a formal category of, the verb.

Even less clearly definable in semantic terms is MOOD. In Latin

and the Romance languages there is the subjunctive mood as

well as the indicative (and the imperative), while Greek has the

optative as well. In so far as the distinction of mood merely

marks another dimension for the classification in formal terms

of the verb forms, it is entirely satisfactory. In Latin we can

To look for some 'real', universal, distinction between mood

and tense, or mood, aspect and tense, is almost certainly

pointless. We shall find that their meanings refer to time, to

possibility, to completion, etc., but seldom will there be any one to- one relation between such meanings and the formal categories.

We should be particularly careful not to take the formal

categories of Latin, assign them generalized meanings and then

impose them, mainly on the basis of those meanings, upon other

languages.

Finally, VOICE raises a few interesting problems. Many languages

show a difference between active and passive with the

object of the active being the subject of the passive in such pairs

of sentences as John saw Bill and Bill was seen by John. This

raises some important theoretical problems that we shall discuss

later, because, unlike tense and mood, voice involves a change

in the position and the function of other words in the sentence.

We often find more than the two distinctions of active and

passive. Greek had a third, the middle, whose meaning was

generally that of doing something for oneself or to oneself. A

more complex, but wholly neat, pattern is found in the Semitic

languages. Consider, for instance, these forms of Tigrinya:

active     qiitiile 'he killed'

passive   tiiqiitle 'he was killed'

causative               'aqti.ile 'he caused to kill'

A different pattern is found in Malagasy (Madagascar), which inaddition to active and passive has a 'circumstantial' voice. Thisallows both indirect objects and instruments to become subjects.There are active and passive sentences that may be translated

as:

'The woman bought the rice for the children with the money.'

'The rice was bought for the children by the woman.'

There are also two possible sentences with the circumstantial

voice:

'The children were bought the rice by the woman.'

'The money was bought-with (used for buying) the rice for

the children.'

Even in English it is not only the object of the active sentence

that may become the subject of the passive. An indirect object

or a 'prepositional' object may also be involved:

The boy was given a book.

This bed has not been slept in.

Moreover, why do we say The bells rang and The fabric washes easily. For we also have active sentences with the bells and the fabric as the objects - They rang the bells and She washed the fabric. It would seem that in the first two sentences rang and washes are somehow passive in meaning, though still active in form. We could, perhaps, treat them as another voice in English, perhaps the English 'middle'; but that would be misleading because there is no distinct set of forms.

Morphology

Grammar is traditionally divided into morphology and syntax. Morphology is essentially the grammar of words and deals with the forms of words, the relation between take and took, dog and dogs.

Syntax is the grammar of sentences and is concerned with the way that sentences are formed (often largely in terms of words ).

Grammar of word (Morphology) & American linguists

many linguists, particularly in America, came to the conclusion that the word was not, or at least not necessarily, the basic unit of grammar, but that often we have to look to something smaller than the word.

This idea is clearly stated by Bloomfield. He pointed out that there are linguistic forms which are never heard in isolation (and so are not words by his definition); these he called bound forms. His examples were the -y Ui/) of Johnny and Billy and the -ing Uirj/) of singing and dancing. These are clearly linguistic forms since they 'are phonetic forms ... with constant meaning' . These forms he called MORPHEMES. His precise definition of a morpheme is 'a linguistic form that bears no partial phonetic-semantic resemblance to any ·other form' . This rather forbidding negative definition is in reality quite simple. By 'bears no partial ... resemblance' he meant that no part of it had any resemblance. To obtain the morphemes, then, we must divide up our linguistic forms until no part of anyone is similar to any other in both its phonetic and its semantic characteristics. Thus dancing cannot be a morpheme because part of it resembles the first part of dances and p~ of it

At first sight the usefulness of the Concept of the morpheme is obvious. We can treat singing and dancing each as two morphemes but with an identical second morpheme -ing, and we can similarly analyse danced and loved or cats and bricks. We must simply divide up the 'complex' forms of language until we arrive at these 'simple' forms (and Bloomfield used the term 'simple' for morphemes).

The remaining task of the grammarian is simply to state

all the possible combinations of these simple forms.

It is obvious that this kind of analysis works admirably for the

agglutinative languages (see p. 52) and that any reasonable

grammars of such languages should be along these lines. Our

Swahili example ·alikuona (p. 51) consists of the elements (i.e.

morphemes) a, Ii, ku and ona. We may well ask, however,

whether such an approach is suitable for the inflectional languages.

These views, and more particularly those of his successors, the 'post-Bloomfieldians', are usually described as 'structuralism', or, strictly, as 'American structuralism'. As the name suggests, the main thesis is that language has a structure. In a general sense, of course, all linguists are structuralists in that they look for regularities, patterns or rules in language. But for this school of linguistics, language structure was of a very specific kind.

Morphemes in general are larger than phonemes; in

fact they are composed of phonemes, it being a requirement that a morpheme must consist of o.ne or mo.re phonemes. Thus singing would be said to. be made up o.f the pho.nemes lsi, Iii, IIJ/, Iiiand IIJI and of the morphemes sing- (o.r Isifj) and -ing (o.r lifj).

To. establish this morphemic and phonemic structure, the linguist must establish first of all what the morphemes and phonemes of the language are by segmenting and classifying actual language material, and then must see clearly what combination of units of the same kind may occur (this is known as 'tactics') and how the morphemes are made up of phonemes.

(1) Bloomfield found it necessary to. talk of 'alternants'. His

example was that of the plural fo.rm in English which even when

written with an -s nevertheless has three pho.no.lo.gical shapes:

lizl in horses

Izi in dogs

lsi in cats

voiceless phonemes (books, cliffs). There is a very similar

situation with the past tense morpheme which has three phonological shapes Itl, Idl and lidl as in liked, loved and hated

respectively (lidl occurring only after It I and Id/). Alternants

of this kind Bloomfield called 'phonetic alternants' because they

'can be described in terms of phonetic modification'. Later

linguists used the term 'allomorph' (or simply 'morph') to

designate the alternants, reserving the term morpheme for the

whole class of alternants. Thus the plural morpheme {s} (with

braces to show that it is a morpheme) would be said to have the

allomorphs lizl, Izl and lsi (with slant lines to show that these

allomorphs consist of phonemes). This particular kind of alternation.

Moreover, was described by them as 'phonologically

determined alternation' since it is determined by the phonological characteristics of the environment. We should perhaps not be surprised that morphemes should undergo 'phonetic modifi~ tion' as Bloomfield called it. It is a very common characteristicof language that one sound seems to determine the nature ofanother adjacent sound. This happens even in agglutinative languages. In Swahili, for instance, m 'him' is replaced by mw

before vowels:

a -Ii - m - penda

a -Ii -mw -ona

'he loved him'

'he saw him'

The introduction of the notion of alternants or allomorphs may seem, then, to be justified, but it carries one important consequence. We can no longer say that morphemes consist. of

phonemes, but rather that the allomorphs or alternants consist

of phonemes. However plausible this modification may be it radically changes the model. (In 3.2 we shall consider a different

model, that deals with this problem in a more satisfying way.)

(2) Bloomfield also noted that there are irregular alternants.

His examples were the first part of:

knife Inaifl knives

mouth Imau61 mouths

house Ihausl houses

Inaiv-zl

Imaua-zl

Ihauz-izl

These are irregular because the final consonant is voiceless in

the singular form and voiced in the plural- If I changes to lvi,

161 to 1M, and lsi to Iz/. This does not occur with most of the

other words that end in the same consonants; we must contrast

cliffs, myths and creases. There are other words with plurals

similar to knives and mouths (e.g. wives, wreaths), but houses is

in this respect quite idiosyncratic: there is no other plural form

in the English language which involves an lsi - Izl alternation

with an lizl ending.

More striking is the p"lural form of ox - oxen. Bloomfield described the ending (/ ani) as 'suppletive' , and it is clear that on

his definition this must be a different morpheme from that

represented by lsi, Izl and liz/. Most later linguists concluded

that it was an allomorph of the same morpheme, and referred to

this kind of alternation as 'morphologically conditioned alternation' because it was conditioned by the occurrence of a particular preceding (or, where appropriate, following) morpheme, and not by any phonological feature. However, with

morphologically conditioned allomorphs, not only is a morpheme

not composed of phonemes (though its alternants may be), but

also there is no simple explanation of the alternation as there is

for the phonologically conditioned allomorphs. It should be

noted, too, that phonological and morphological conditioning

are not alternatives. Many morphemes are both morphologically

and phonologically conditioned. Thus the occurrence of lizl is

morphologically conditioned in that it occurs with glass, horse,

crease, etc., but not with ox, and phonologically conditioned in

terms of the preceding sibilant or affricate.