Perspectives of the Analysis of Syntax

Also Read

1.0 Introduction

Having delimited the domain of syntactic investigation in the previous unit, we may now proceed to discuss the ways in which words or strings of words are organized within the sentence. Within the linguistic tradition, it is possible, with a certain amount of abstraction, to discern two distinct trends, which may loosely be called the 'bottom-up' or word relation approach, and the 'top-down' or constituent approach. The two approaches do not meet in the middle, but both have things of value to contribute to our understanding of syntactic structure. We shall discuss the bottom-up approach first, then contrast it with the top-down approach, which supplanted it as the dominant trend in syntax around the middle of the last century.

2.0 Objectives

At the end of the unit, you will be expected to:

- Explain what bottom-up and top-down approaches mean and how they work.

- Discuss the contributions of these approaches in our knowledge of syntactic structure.

- Highlight their limitations in syntactic analysis.

3.0 Main Content

3.1 'Bottom-Up' or 'Morpheme-to-Utterance' Approach (Word Relation Approach)

There are several methods of syntactic analysis. Two of the methods we shall discuss here involve searching for immediate constituents within utterances by comparing samples. One is by identifying the immediate constituents moving from the morpheme to the utterance (popularly called 'bottom-up' approach). In proceeding from morpheme to utterance, constituents cuts are progressively erased, forming larger and larger constituents back to the utterance itself.

The bottom-up approach starts from the lower end of the domain of syntax – i.e. with words – and attempts to specify ways in which they may be combined and in which they relate to each other. The bottom-up approach was largely developed by the American linguist, Zelling Harris.

The first thing to be done in this approach is to establish classes of units at a certain level so that their occurrence with respect to each other may be specified. An example of such classification is the traditional 'parts of speech' into which most of the words of English are grouped. (A detailed discussion of 'parts of speech' will be undertaken in unit 5). The following eight parts of speech were established by the traditional grammarians: Noun (the name of a person, place or thing), Verb (a word that expresses an action or state of being), Adjective (a word used to modify a noun), Pronoun (a word that substitutes for a noun), Adverb (a word used to modify a verb, adjective or other adverbs), Preposition (a word that links a noun or pronoun to some other parts of the sentenced), Conjunction (a word used for joining one sentence to another, or one word to another of the same or similar parts of speech), and Interjections (exclamatory sounds that express some feeling of the mind). To find out the 'part of speech' to which a word belongs, ask yourself: 'What kind of work does the word do in the sentence?'; 'what part does it play in helping to make the sentence?' If the word gives a name to an individual, person or thing, or to some kind of person or thing, the word is a Noun. If the word refers to some person or thing without giving a name to the person or thing referred to, the word is a pronoun. This was how traditional grammarians classified words.

In spite of the inadequacies of traditional parts of speech, it is a remarkably effective tool for the analysis of English syntactic structure. It performs two essential tasks. On the one hand, it provides a framework into which the vast majority of English words may be fitted. On the other, it is an essential prerequisite for the statement of possible word combinations in English. These may be formulated in quite a traditional way, along the following lines:

i. Adverbs precede the adjectives in English (e.g. horribly inadequate).

ii. Adjectives precede the noun and follow the article in English (e.g. the big picture).

iii. English sentences have the order noun-verb-noun (e.g. John likes yam).

Such generalizations seem obvious to us, but they presuppose a surprisingly large amount of linguistic analysis, and it is only when such straightforward principles of grammar are established that it is possible to move on to consider the more elaborate deviations from these patterns which many consider to be the more interesting part of syntax.

Having established the units, the linguist using 'bottom-up' approach begins to combine morphemes into phrases and then phrases into sentences or utterances. For example, the sentence:

1. Those three boys ate the food.

will be analyzed as follows:

{ate} (actually {eat} + {past}) {three} + {boys} (actually {boy} + {plural}) {food} {those} + {three} + {boys} {ate} + {the} + {food} {those} + {three} {boys} + {ate} + {the} + {food}

This can be represented in another way:

1. Those /three/boys/ate/the/food 2. Those/three boys/ate/the food 3. Those three boys/ate/the food 4. Those three boys/ate the food 5. Those three boys ate the food

The end result after combining the morphemes progressively as they 'go together' is a distinct unit, an utterance, or a sentence.

In the discussion above, we have tried to show how 'bottom-up' or 'morpheme-to-utterance' approach works. First, we establish the word class of each element. Next, we establish the ways in which elements combine and relate to each other until we get to the sentence level. We find that this approach is capable of providing structural characterizations of a large number of English sentences and, at the same time, of specifying as un-English a large number of possible combinations of English utterances.

SELF-ASSESSMENT EXERCISE 1

1. What is 'bottom-up' approach in syntactic analysis?

2. State the step-by-step procedures of 'bottom-up' approach in syntactic analysis.

3.2 'Top-Down' or Constituent Structure Approach

We have tried to present the general outlines of the 'bottom-up' approach to syntactic analysis – one which is firmly rooted in centuries of use in the European grammatical tradition. But there are good reasons to be unhappy about some aspects of such an approach. The definition of the parts of speech is one problematic area: it is very easy to find fault (in principle, rather than merely in detail) with the semantic parts of the definitions. Nouns, for instance, are defined in terms of what are essentially physical objects, but it is far from difficult to find nouns which denote other than physical objects; indeed, it is virtually impossible to find concepts expressed by verbs, adjectives, or adverbs which cannot also be expressed by nouns, whether directly derived from other words (e.g. be, - being; red-redness) or not (e.g. hit – blow, big – size). If nouns and verbs were semantically quite distinct, then it would be contradictory for there to be such things as 'verbal nouns' (i.e. gerunds – liking, wanting, etc.). The same sort of objection may be made to all the other semantic definitions as well. Although there is some sense in these notional definitions, but as criteria for determining the membership of a category, they are undoubtedly inadequate.Misgivings also arise as regards the grammatical relations which form an essential part of bottom-up grammatical analysis. The semantic parts of such definitions suffer from the same faults as do those of the parts of speech. For example, the definition of the subject of a verb as actor, experiencer, or possessor of a property is not obviously satisfied by this cupboard in (1) below:

1. This cupboard contains dangerous medicines.

Even the usual definitions of subject and object don’t seem quite appropriate for (2) and (3):

2. Samson suffered a stroke.

3. This problem taxed many of the greatest minds of the time.

But even assuming the adequacy in general of these definitions (and there is no doubt that they cover the vast majority of the cases), there are further problems. Some word relations do not so obviously fit into the head – governed term pattern. For example, the phrase many men give grounds for claiming either word to be the head. Either of the words may occur independently as the subject of a verb:

4. Men were killed.

5. Many were killed.

Whenever either word occurs, a word of the other class could occur alongside it; men suggest a quantity, which may be specified, like suggests a set of objects, which also may be specified. Nor does the relation seem to be way different from that which obtains in combinations such as many of the men, which consists of two phrases, many and of the men; of these many is obligatory, but of the men is not. It is traditional to treat quantity words (quantifiers) such as many on a par with adjectives, and to analyze the noun as the head of the construction, but there seems to be no sound argument other than the force of tradition which requires us to analyze it in this way.

Finally, there are constructions which fit rather badly into an approach which describes syntactic structure in terms of relations between individual words. Take the following examples:

6. Tom and Jane walked up the street.

7. Mary has been eating apples.

In (6) it seems obvious that Tom and Jane as a whole is the subject of the verb walked: but establishing what could be the head and corresponding relations within that three-word phrase seems hardly possible.

Such problems, and especially the difficulty of applying the traditional approach to languages of widely differing structure, led naturally to the development of what we have called the ‘top-down’ approach. A simple and appealing version of such an approach was developed in the American structuralist tradition under the name of ‘immediate constituent analysis’. Essentially this approach relies on the possibility of intersubstitution of strings of words or morphemes across different sentences, a procedure which is taken as establishing their equivalence as syntactic units.

The idea here is that sentences are constructed from groups of words, often paired rather than from single words added one onto the next. These groups of words in turn cluster with other groups, layer upon layer of word pairs and pair groups, which eventually build a sentence. When presented with a sentence such as (8) below, a native speaker will find little difficulty in segmenting it into significant ‘chunks’.

8. The man that I saw was wearing a blue jacket.

It is intuitively rather obvious that strings of words from (8) such as the man that I saw, a blue jacket, the man, that I saw, etc. are rather different in status from strings such as I saw was, man that, wearing a blue, etc.

A three-word sentence such as (9) admits of three possible segmentations (assuming continuity of resulting segments).

9. Peter ran away.

We may treat it as involving three isolated words as in (10), or we may decide that an adjacent pair of words is more closely related than either are to the third word as in (11) and (12).

10. [Peter] [ran] [away].

11. [Peter ran] [away].

12. [Peter] [ran away].

The criterion that we will accept as choosing between these segmentations is that of regularity of substitutability. So another three-word sentence, such as (13), which allows the same segmentation, allows free substitution of segments only with (12).

13. Mary likes custard.

In other words, substituting away for custard (or vice versa) yields an unacceptable sentence:

14. Mary likes away.

15. Peter ran custard.

But Peter and Mary on the one hand, and ran away and likes custard on the other, are freely inter-substitutable:

16. Peter likes custard.

17. Mary ran away.

This type of segmentation is applicable to other sentences. Let us consider the sentence:

IC analysis also demonstrates two important points about English syntax. The first reinforces what we already knew from using test frames: English syntax is highly positional in structure – English is a word-order language – and words placed next to each other are usually semantically connected. The second point is that groups of words in English do indeed function as single units of syntax.

We have seen that the optimal result of immediate constituent analysis for typical sentences of English reveals a consistent pattern in all of them; the normal constituent break in a full sentence of English are two constituents which correspond fairly closely to the traditional notions of subject and predicate. Another observation which was central to the development of this type of analysis was that constituents which result from such a procedure tend to behave syntactically in ways which are similar to the behavior of individual words. A constituent of whatever length is then said to be an expansion of the word class with similar behavior. This notion of expansion is directly related to that of distribution. If two strings of words are virtually replaceable in all contexts, they are said to have distribution.

4.0 Conclusion

In this unit, we discussed two perspectives of the analysis of syntax: the ‘bottom-up’ and the ‘top-down’ approaches. As we have seen, each approach treats a sentence as a sequence of words, each word bearing a relation of some sort to at least one other word of the sequence. The other approach involves breaking a sentence down into chunks, and continuing this process of analysis until all the chunks into which it is analyzed are words.

5.0 Summary

You have learnt in this unit,

- What ‘bottom-up’ approach is and how it works.

- What ‘top-down’ approach is and how it works.

6.0 Tutor-Marked Assignment

Use any of the approaches discussed above in analyzing the following sentences. Explain your step-by-step procedure.

1. The heavy rains in Benin City have rendered many streets impassable.

2. The militants have been making life intolerable for city residents.

3. They played a wonderful game.

7.0 REFERENCES/FURTHER READINGS

Atkinson, Martin; David Kilby and Iggy Roca (1982). Foundations of

General Linguistics. London: George Allen: Unwin.

Clark, Virginia P., Paul A. Eschholz and Alfred F. Rosa. (Eds)

Language: An Introductory Readings 3rd Edition. New York: St.

Martins Press.