Bold type 5 страница

As we shall see in Part 3, the relation between lexical and grammatical meaning varies from language, to language: what is encoded lexically (lexicalized) in one language may be encoded grammatically (grammaticalized) in another. The grammaticalization of meaning, as we shall also see later, is not simply, or primarily, a matter of inflection (even in languages which, unlike English, have a very rich inflectional system). Far

 

2.1 Forms and expressions 53

more important are the syntactic differences between one grammatical construction and another.

At this point, however, it may be noted that, when word-forms are considered, not just as forms, but as forms invested with grammatical meaning, yet another sense both of 'form' and of 'word' comes to light. Consider, for example, the follow­ing sentences:

(4) 'That sheep over there belongs to the farmer next door'

(5) 'Those sheep over there belong to the farmer next door'.

Is the second word-form of (4) the same as the second word-form of (5)? The distinction that we have drawn between forms and expressions does not, of itself, suffice to answer the question in a case like this. Let us grant immediately that the two word-forms are identical in respect both of their phonological form (in the spoken language) and of their orthographic form (in the written language): they are formally identical. But they are not grammatically identical. Whether we say that the second word-form of (4) is the same as the second word-form of (5) depends, therefore, on whether, in putting this question, we are concerned with formal identity alone - phonological or ortho­graphic, as the case may be - or with both formal and gramma­tical identity. The two word-forms that occur in the second position of (4) and (5) are formally identical, but grammatically distinct, forms of the same lexeme. More precisely, they are inflectionally, or morphosyntactically, distinct forms of the same lexeme. The way in which this phenomenon is handled by grammarians will differ according to the model of grammar which they adopt.

What has been said in this section about Peirce's type/token distinction, about the different senses in which 'word' is used both technically and non-technically in linguistics, about the distinction between forms and expressions and about lexical and grammatical meaning is sufficient for the time being. It may seem at first sight that, in this section, I have been unneces­sarily pedantic in my regimentation and extension of the every­day metalanguage. This is not so. Whatever terms we use to draw the distinctions that have been drawn here, the distinctions

54 Words as meaningful units

themselves must be drawn if we are to avoid the confusion and equivocation which is almost inevitably associated with what I referred to in the Preface as the pseudo-simplicity of so-called plain English.

All the points that I have made could be developed at great length, and would need to be. in a fuller account of what is com­monly, but imprecisely, referred to as word-meaning. They would also need to be formulated somewhat differently in rela­tion to particular theories of phonology, syntax and morphol­ogy. I have deliberately adopted a rather traditional view of the grammatical and lexical structure of languages. There are two reasons why I have done this. The first is that this view is the one that is reflected in the most widely used authoritative dictionaries and reference grammars of English and other languages and is also the view that is taught or taken for granted in most schools: it may therefore be assumed to be a view which is familiar to most readers of this book (even if they are riot in command of all the technical terminology). The second reason is that, although various refinements and qualifications have to be made to this traditional view in the light of developments in modern grammatical theory, so-called traditional grammar 'with the necessary refinements and qualifications to which I have referred) serves better than any alternative so far available as an established standard system into which and out of which other competing systems can be translated. Students who have already had some training in modern grammatical theory will find it instructive to carry out this exercise in translation from one metalanguage to another as we proceed through this and the following chapters.

 

2.2 HOMONYMY AND POLYSEMY; LEXICAL AND GRAMMATICAL AMBIGUITY

 

What is traditionally described as homonymy was illustrated in Chapter 1 by means of the no less traditional examples of 'bank₁' and 'bank₂', the former meaning "financial institution" and the latter "sloping side of a river". The examples are

 

2.2 Homonymy and polysemy; lexical and grammatical ambiguity 55

appropriate enough. But the traditional definition of homonymy is, to say the least, imprecise.

Homonyms are traditionally defined as different words with the same form. We can immediately improve this definition, in the light of what was said in the preceding section, by substitut­ing 'lexeme' for 'word'. But the definition is still defective in that it fails to take account of the fact that, in many languages, most lexemes have not one, but several, forms. Also, it says noth­ing about grammatical equivalence.

Let us begin, therefore, by establishing a notion of absolute homonymy. Absolute homonyms will satisfy the following three conditions (in addition to the necessary minimal condition for all kinds of homonymy — identity of at least one form):

(i) they will be unrelated in meaning; (ii) all their forms will be identical;

(iii) the identical forms will be grammatically equivalent.

Absolute homonymy is common enough: cf. 'bank₁', 'bank₂'; 'sole₁' ("bottom of foot or shoe"), 'sole₂' ("kind of fish"); etc.

But there are also many different kinds of what I will call par­tial homonymy: i.e., cases where (a) there is identity of (mini­mally) one form and (b) one or two, but not all three, of the above conditions are satisfied. For example, the verbs 'find' and 'found' share the form found, but not finds, fading, or founds, found­ing, etc.; and found as a form of 'find' is not grammatically equivalent to found as a form of 'found'. In this case, as generally in English, the failure to satisfy (ii) correlates with the failure to satisfy (iii). However, it is important to realize that the last two conditions of absolute homonymy made explicit in the previous paragraph are logically independent. They are usually taken for granted without discussion in traditional accounts of the topic.

It is particularly important to note the condition of gramma­tical equivalence, and the fact that this is a matter of degree. Although found as a form of 'find' is not grammatically equiva­lent to found as a form of 'found', it is in both cases a transitive-verb form. Consequently, there are certain contexts in which

 

56 Words as meaningful units

found may be construed, grammatically, either as a form of 'found' or as form of 'find'. For example (see (3) in section 2.1):

(6) They found hospitals and charitable institutions

can be construed as a present-tense sentence containing a form of the verb 'found' or. alternatively, as a past-tense utterance con­taining a form of 'find'. The fact that 'found' and 'find' are tran­sitive verbs - and to this degree (though not fully) grammatically equivalent - means that they can both take a noun-phrase such as 'hospitals and charitable institutions' as their direct object. And since 'hospitals and charitable institu­tions' is, not only grammatically, but also semantically, accept­able as the direct object of both verbs, (6) is ambiguous.

The ambiguity of (6) is partly lexical and partly grammatical. It is lexically ambiguous in so far as its ambiguity depends upon a difference in the lexical meaning of the two partially homon-ymous lexemes 'found' and 'find'. It is grammatically ambigu­ous in so far as its ambiguity depends upon the (semantically relevant) grammatical non-equivalence of found construed as a form of'found' and of found construed as a form of 'find'.

The reason why it is important for the semanticist to take note of grammatical equivalence, is that in general, it is this which determines whether homonymy (absolute or partial as the case may be) results in ambiguity. If have is inserted before found in (6), to yield

(7) They have found hospitals and charitable institutions,

the ambiguity disappears. The effect of putting the form have before the form found is to change the morphosyntactic identity of the latter: on the assumption that (7) is indeed fully gramma­tical in English, found must now be construed as a past participle. The past-participle form of 'find' happens to be formally identi­cal with the past-tense form of 'find' (both phonologically and orthographically). The past-participle form of 'found', on the other hand, is formally identical with its past-tense form: founded. (In this respect, 'found' is like most other English verbs; 'find', in contrast, belongs to a particular subclass of what are tradi­tionally described as irregular, or strong, verbs.)

2.2 Homonymy and polysemy; lexical and grammatical ambiguity 57

The ambiguity that is manifest in (6) also disappears if he or she is substituted for they:

(8) He/she-found hospitals and charitable institutions.

The reason now is that in English, whereas there is formal iden­tity (except for the verb 'be') between singular and plural forms in all simple past-tense verb-forms, what are traditionally called third-person singular and plural forms are formally distinct in the simple present tense of the indicative (in all verbs other than modals, such as 'may' or 'can'): cf. finds: find and founds: found. It follows that in (8) found must be construed as a form of 'find' and therefore as a past-tense form. To be contrasted with (6) are, on the one hand,

(9) He/she founds hospitals and other charitable institutions
and, on the other,

(10) He/she founded hospitals and other charitable institutions.

Ambiguity which results from absolute homonymy cannot be eliminated by manipulating the grammatical environment in this way. But, it is quite possible for the partial homonymy of two lexemes rarely or never to result in ambiguity: ambiguity is forestalled, as it were, if the shared forms are prohibited from. occurring in the same grammatical environments. For example, the partial homonymy of the adjective 'last₁' (as in 'last₁ week') and the verb 'last₂' (as in 'Bricks last a long time') rarely pro­duces ambiguity. Their sole shared form, last, is almost always readily identifiable as a form of the one or the other by virtue of the grammatical environment in which it occurs.

We shall return to the question of ambiguity in a later chap­ter. We shall then see that the kind of grammatical ambiguity (combined with lexical ambiguity) which has been exemplified here in connexion with the traditional notion of homonymy is just one kind of grammatical ambiguity. It has been mentioned at this point because many general accounts of homonymy, both traditional and modern, fail to draw attention to the complexity and variety of the grammatical conditions that must be satisfied if partial homonymy is to result in ambiguity.

 

 

58 Words as meaningful units

Many accounts of homonymy also fail to point out that partial homonymy does not necessarily involve identity of either the citation-forms or the underlying base-forms of the lexemes in question. For example, the noun 'rung' and the verb 'ring' are partial homonyms:

(11) A rung of the ladder was broken;

(12) The bell was rung at midnight.

The reason why this kind of partial homonymy is 6ften not recognized in standard treatments, traditional or modern, is that the former tend to concentrate on citation-forms, whereas the latter frequently restrict their discussion of homonymy to base-forms. It so happens, of course, that in English the citation-form coincides with the base-form in all morphologi­cally regular lexemes. But this is not so in all languages, as far as the traditional ordinary-language citation-forms of lexemes are concerned. For the semanticist, as we have seen, the question at issue is whether and to what degree homonymy produces ambi­guity. From this point of view there is nothing special about either citation-forms or base-forms.

Let us now turn to polysemy. Whereas homonymy (whether absolute or partial) is a relation that holds between two or more distinct lexemes, polysemy ("multiple meaning") is a property of single lexemes. This is how the distinction is traditionally drawn. But everyone who draws this distinction also recognizes that the difference between homonymy and polysemy is not always clear-cut in particular instances It has been demon­strated, for English, that there is a good deal of agreement among native speakers, in most cases, as to what counts as the one and what counts as the other. But there arc also very many instances about which native speakers will hesitate or be in dis­agreement. What, then, is the difference in theory between homonymy and polysemy?

The two criteria that are usually invoked in this connexion have already been mentioned in Chapter 1: etymology (the his­torical source of the words) and relatedness of meaning. In general, the etymological criterion supports the native speaker's untutored intuitions about particular lexemes. For example,

2.2 Homonymy and polysemy; lexical and grammatical ambiguity 59

most native speakers of English would probably classify 'bat₁' ("furry mammal with membranous wings") and 'bat₂' ("imple­ment for striking a ball in certain games") as different lexemes; and these two words do indeed differ in respect of their historical source, 'bat₁' being derived from a regional variant of Middle English 'bakke', and 'bat₂' from Old English 'batt' meaning "club, cudgel".

To say that Etymology generally supports the intuitions of native speakers is not to say that this is always the case. It some­times happens that lexemes which the average speaker of the language thinks of as being semantically unrelated have come from the same source. The homonyms 'sole₁' ("bottom of foot or shoe") and 'sole₂' ("kind offish"), which I mentioned above, constitute a much-quoted example; and there are others, no less striking, to be found in the handbooks. Less common is the con­verse situation where historically unrelated meanings are per­ceived by native speakers as having the same kind of connexion as the distinguishable meanings of a single polysemous lexeme. But there are several examples of what, from a historical point of view, is quite clearly homonymy being reinterpreted by later generations of speakers as polysemy. It falls within the scope of what is commonly referred to by linguists as popular etymol­ogy. Today, for example, a number of speakers assume that 'shock₁' as in 'shock of corn' is the same as 'shock₂' as in 'shock of hair'. Yet historically, they have different origins.

There are exceptions, then, of both kinds. Nevertheless, the generalization that I have just made is undoubtedly correct: in most cases, etymology supports the average native speaker's intuitive sense of the distinction between homonymy and poly­semy. And we shall see presently that there are good reasons why this should be so. One of the principal factors operative in semantic change is metaphorical extension, as when 'foot' mean­ing "terminal part of a leg" also came to mean "lowest part of a hill or mountain". And it is metaphorical extension as a syn-chronic process that is at issue when one refers to the related meanings of polysemous lexemes. There are, of course, other kinds of relatedness of meaning which are relevant in this con­nexion. But metaphorical creativity (in the broadest sense of

 

 

60 Words as meaningful units

'metaphorical' is part of everyone's linguistic competence. In the last resort, it is impossible to draw a sharp distinction between the spontaneous extension or transfer of meaning by individual speakers on particular occasions and their use of the pre-existing, or institutionalized, extended and transferred meanings of a lexeme that are to be found in a dictionary. This fact has important implications for linguistic theory that go way beyond the traditional, and perhaps insoluble, prob­lem of distinguishing polysemy from homonymy.

2.3 SYNONYMY

Expressions with the same meaning are synonymous. Two points should be noted about this definition. First it does not restrict the relation of synonymy to lexemes: it allows for the pos­sibility that lexically simple expressions may have the same meaning as lexically complex expressions. Second, it makes identity, not merely similarity, of meaning the criterion of syn­onymy.

In this latter respect, it differs from the definition of synonymy that will be found in many standard dictionaries and the one with which lexicographers themselves customarily operate. Many of the expressions listed as synonymous in ordinary or spe­cialized dictionaries (including Roget's Thesaurus and other dic­tionaries of synonyms and antonyms) are what may be called near-synonyms: expressions that are more or less similar, but not identical, in meaning. Near-synonymy, as we shall see, is not to be confused with various kinds of what I will call partial synonymy, which meet the criterion of identity of meaning, but which, for various reasons, fail to meet the conditions of what is generally referred to as absolute synonymy. Typical examples of near-synonyms in English are 'mist' arid 'fog', 'stream' and 'brook', and 'dive' and 'pfunge'.

Let me now introduce the notion of absolute synonymy, in contrast not only with near-synonymy, but also with the broader notion of synonymy, just defined, which covers both absolute and partial (i.e.. non-absolute) synonymy. It is by now almost a

2.3 Synonymy 61

truism that absolute synonymy is extremely rare - at least as a relation between lexemes — in natural languages. (It is not rare of course as a relation between lexically composite expressions.) Two (or more) expressions are absolutely synonymous if, and only if, they satisfy the following three conditions:

(i) all their meanings are identical; (ii) they are synonymous in all contexts; (iii) they are semantically equivalent (i.e., their meaning or

meanings are identical) on all dimensions of meaning,

descriptive and non-descriptive.

Although one or more of these conditions are commonly men­tioned in the literature, in discussions of absolute synonymy, it is seldom pointed out that they are logically independent of one another; and non-absolute, or partial, synonymy is not always clearly distinguished from near-synonymy.

This being so, I wish to insist upon the importance of: (a) not confusing near-synonyms with partial synonyms; and (b) not making the assumption that failure to satisfy one of the condi­tions of absolute synonymy necessarily involves the failure to satisfy either or both of the other conditions. Let us take each of the conditions of absolute synonymy in turn.

Standard dictionaries of English treat the adjectives 'big' and 'large' as polyscmous (though they vary in the number of mean­ings that they assign to each). In one of their meanings, exempli­fied by

(13) 'They live in a big/large house',

the two words would generally be regarded as synonymous. It is easy to show, however, that 'big' and 'large' arc not synonymous in all of their meanings: i.e., that they fail to satisfy condition (i) and so are only partially, not absolutely, synonymous.'The fol­lowing sentence,

(14) 'I will tell my big sister',

is lexically ambiguous, by virtue of the polysemy of 'big', in a way that

 

62 Words as meaningful units

15) 'I will tell my large sister'

is not. All three sentences are well formfed and intcrpretable. They show that 'big' has at least one meaning which it docs not share with 'large'. There are many such examples of polysemous lexemes that are synonymous in One or more, but not all, of their meanings.

Let us now turn to condition (ii). What is at issue here is the

collocational range of an expression: the set of contexts in which it can occur (its collocations). It might be thought that the collocational range of an expression is wholly determined by its meaning, so that synonyms must of necessity have the same collocational range. But this does not seem to be so. Once again, 'big' and 'large' will serve as an example. There are many contexts in which 'large' cannot be substituted for 'big' (in the meaning which 'big' shares with 'large') without violat­ing the collocational restrictions of the one or the other. For example, 'large' is not interchangeable with 'big' in

(16) 'You are making a big mistake'.

The sentence

(17) 'You are making a large mistake'

is. presumably, not only grammatically well-formed, but also meaningful. It is however collocationally unacceptable or un-idiomatic. And yet 'big' seems to have the same meaning in (16) as it does in phrases such as 'a big house', for which we could, as we have seen, substitute 'a large house'.

It is tempting to argue, in cases like this, that there must be some subtle difference of lexical meaning which accounts for the collocational differences, such that it is not synonymy, but near-synonymy, that is involved. Very often, undoubtedly, col­locational differences can be satisfactorily explained, in terms of independently ascertainable differences of meaning. But this is not always so. We must be careful therefore not to assume that the collocational range of a lexeme is predictable from its mean­ing. Indeed, there are cases where it can be argued that the collo­cations of a lexeme are part of its meaning. This, regrettably, is

2.3 Synonymy 63

 

one of many aspects of lexical semantics that cannot be dealt
with in this book.

The third of the conditions of absolute synonymy listed above was identity on all dimensions of meaning. The most widely recognized dimension, of meaning that is relevant to this condi-tion is descriptive (or prepositional) meaning (see section 1.7). In fact, many theories of semantics would restrict the notion of synonymy to what I will call descriptive synonymy: identity of descriptive meaning. What precisely is meant by identity of descriptive meaning is a question that will be taken up in Part 3. For the present, it will be sufficient to say that two expressions have the same descriptive meaning (i.e., are descriptively synon­ymous) if, and only if, propositions containing the one necess­arily imply otherwise identical propositions containing the other, and vice versa. By this criterion (which will be reformu­lated in Part 3 in terms of the truth-conditional equivalence of sentences), 'big' and 'large' are descriptively synonymous (in one of their meanings and over a certain range of contexts). For instance, one cannot without contradiction simultaneously assert that someone lives in a big house and deny that they live in a large house.

One of the classic examples of descriptive synonymy is the relation that holds (or perhaps used to hold) in English between 'bachelor' (in one of the meanings of 'bachelor') and 'unmarried man'. (There are those who would deny that these two expres­sions are descriptively synonymous, nowadays, on the grounds that a divorced man, though unmarried, is not a bachelor. The point is debatable; and, since it can be exploited for more general theoretical purposes, I will return to it in a later chapter. But the principle that the example is intended to illustrate is clear enough.) One tests for descriptive synonymy, in this case, by investigating whether anyone truly, or correctly, described as a bachelor is truly describable as an unmarried man, and vice versa. It may well be that for some speakers the expressions are synonymous and for others they are not, and that for a third group the situation is unclear. (Those who hold that 'unmarried' means, not simply "not married", but "never having been mar­ried", and cannot be correctly applied to divorcees – together

 

64 Words as meaningful units

with those, if any. who would readily apply both 'bachelor' and 'unmarried' to divorcees - will presumably treat 'bachelor" and 'unmarried man' as descriptively synonymous.)

When it comes to expressive (or, socio-exprqssivc) mean­ing - and this is the only kind of non-descriptive meaning that we will take into account here - there is no readily available and reasonably objective criterion which enables us to decide between identity and difference. But it is none the less possible, in particular instances, to determine that two or more descrip­tively synonymous expressions differ in respect of the degree or nature of their expressive meaning. For example, it is intuitively obvious that a whole set of words including 'huge', 'enormous', 'gigantic' and 'colossal' are more expressive of their speakers' feelings towards what they are describing than 'very big' or 'very large', with which they are perhaps descriptively synon­ymous. It is more difficult to compare 'huge', 'enormous', 'gigantic' and 'colossal' among themselves in terms of their degree of expressivity. But speakers may have clear intuitions about two or more of them; and the question is, in principle, decidable by means of relatively objective psychological tests.

As to expressions which differ in the nature of their expressive meaning, the most obvious difference is between those which imply approval or disapproval and those which are neutral with respect to expressivity.'Textbooks' of linguistic semantics are full of examples, such as 'statesman' versus 'politician', 'thrifty' versus 'mean', 'stingy' versus 'economical', 'stink' versus 'stench' versus 'fragrance' versus 'smell', 'crafty' versus 'cun­ning' versus 'skilful' versus 'clever', and so on. In many cases, the fact that an expression implies approval or disapproval is much more readily ascertainable than is its descriptive meaning (if it has any). This is true, for example, of words such as 'bitch' or 'swine' used in what was once, but is perhaps no longer for most speakers of English, their metaphorical sense. Under what conditions can one truly describe a person as a bitch or swine? In cases like this it is surely the expressive, rather than the descriptive, component of meaning that is dominant.

Most of the lexemes in everyday use have both a descriptive and an expressive meaning. Indeed, as certain philosophers of

 

2.4 Full and empty word-forms 65

language have pointed out in respect of the vocabulary of moral and aesthetic statements, it may be even theoretically impossible at times to separate the descriptive from the expressive. How­ever that may be, knowing the expressive (or socio-expressive) meaning of a lexeme is just as much part of one's competence in a language as knowing its descriptive meaning. This point should be constantly borne in mind throughout this book, even though we shall be concerned almost, exclusively with descrip-tive meaning in our discussion of lexical structure in Chapter 3 and in several of the later chapters.

Synonymy has been discussed and richly exemplified from many points of view, not only in works devoted to linguistic semantics as such, but also in handbooks of stylistics, rhetoric and literary criticism. My main purpose, in the brief account that has been given here, has been to emphasize the theoretical importance of distinguishing the several kinds of partial, or non-absolute, synonymy from one another and from near-synonymy. In doing so, I have been obliged to gloss over a num­ber of difficulties and complications that a more comprehensive discussion of synonymy would require us to deal with. Some of these will be mentioned in Chapter 4, as far as descriptive syn­onymy is concerned, in connexion with the notion of entailment.