Bold type 6 страница

2.4 FULL AND EMPTY WORD-FORMS

The word-forms of English, like the word-forms of many lan­guages, can be put into two classes. One class consists of full forms such as man, came, green, badly; the other of empty forms such as the, of, and, to, if. The distinction between the two classes is not always clear-cut. But it is intuitively recognizable in the examples that I have just given. And it has been drawn on non-intuitive grounds by grammarians, by applying a variety of cri­teria. Essentially the same distinction was drawn, centuries ago, in the Chinese grammatical tradition; at the end of the nine­teenth century, by the English grammarian Henry Sweet; and at the height of post-Bloomfieldian structuralism, in the 1950s, by the American linguist C. C. Fries (1952). It subsequently found its way into many of the textbooks of applied linguistics

 

66 Words as meaningful units

and practical teaching-grammars of English and other languages in the period preceding the rise of Chomskyan gen­erative grammar in the 1960s. It correlates with the distinction between open-class and closed-class word-forms which is drawn (in these or other terms) in many modern schools of grammatical theory.

The terms that I have chosen, taken from the Chinese tradi­tion, emphasize the intuitively evident semantic difference between typical members of one class and typical members of the other. Empty word-forms may not be entirely devoid of meaning (though some of them are in certain contexts). But, in an intuitively clear sense of 'meaningful', they are generally less meaningful than full word-forms are: they are more easily pre­dictable in the contexts in which they occur. Hence their omis­sion in headlines, telegrams, etc., and perhaps also in the utterances of very young children as they pass through early stages of language-acquisition. Full word-forms in English are forms of the major parts of speech, such as nouns, verbs and adjectives; empty word-forms (in languages that have them) belong to a wide variety of classes - such as prepositions, definite and indefinite articles, conjunctions, and certain pronouns and adverbs - which combine with the major parts of speech in grammatically well-formed phrases and sentences and which. (unlike the major parts of speech) tend to be defined mainly in terms of their syntactic function, rather than semantically.

Other terms found in the literature, more or less equivalent to 'empty word-form', are 'form word', 'function word', 'gramma­tical word' and 'structural word'. All these terms reflect the view that what I am here calling empty word-forms differ gram­matically and semantically from full word-forms. They are usually defined within the framework of Bloomfieldian and post-Bloomfieldian (including Chomskyan) mdrpheme-based grammar on the basis of Bloomfield's definition of the word (in the sense of 'word-form') as a minimal free form. We are operat­ing throughout this book within the more traditional framework of what has been called word-and-paradigm grammar. But what I have to say here, and indeed throughout this book, could be reformulated without difficulty in the

2.4 Full and empty word-forms 67

terminology of any of several different schools of grammatical theory, old and new, and is intended to be, as far as possible, theory-neutral. I have chosen to use 'empty word-form' and 'full word-form' because these terms emphasize the semantic dimension of the difference between the two classes.

Looked at from a grammatical point of view, empty word-forms can be seen as playing much the same role in non-inflecting, or lowly inflecting, languages as do prefixes, suffixes, etc. in highly inflecting languages. For example, a prepositional phrase such as to John when it occurs in the indirect-object posi­tion after the verb 'give' in English can be matched, semantically and grammatically, in many highly inflecting languages, such as Latin or Russian (and many other languages belonging to many different language-families throughout the world), with what is traditionally referred to as the dative (or allative) form of the noun, which contrasts with other syntactically and/or semantically distinct forms of the same lexeme in having the dative (or allative) suffix, rather than the nominative, accus­ative, genitive, etc. suffix, attached to the base-form. Similarly, for the definite article the. The vast majority of the languages of the world do not have a separate word-form which can be identi­fied grammatically and semantically with the English definite article.

Indeed, most natural languages do not encode a category of definiteness as such at all, either grammatically or lexically. Some languages which do encode definiteness (in so far as this is identifiable and separable from other semantic categories across languages) do so inflectionally, in much the same way that the indirect-object function is expressed inflectionally by the dative case in Latin. In view of the attention that twentieth-century English-speaking logicians, beginning with Russell (1905), have given to the analysis of noun-phrases containing the defi­nite article, it is worth noting the non-universality, not just of the definite article, but also of anything that might be called a semantic category of definiteness, in natural languages. But this is an issue which does not concern us for the moment. I have mentioned the English definite article at this point as an example of the class of what I am calling empty word-forms.

68 Words as meaningful units

It will be noted that, although I have referred to empty word-forms as word-forms, I have not said that they are forms of lexemes (as dog is a form of 'dog', ran is a form of 'run', and so on).It is a moot point whether forms such as the or to (in its indirect object function or its infinitive-forming function at least) should be listed in the dictionary of a language or accounted for within the grammar. This is an issue which cannot be settled except within the framework of one grammatical theory or another. But whatever view is taken on this issue, the main point to be made here is that, even if they are listed in dictionaries of the language (whether for reasons of practical convenience or on the basis of a theoretically defensible notion of the distinction between grammar and lexicon), empty 'word-forms, such as the, of, and, to and if in English, are hot fully lexical. They may be words in the sense of 'word-form', but ₍they are not words in the full sense.

Not only do empty word-forms tend to be less meaningful than full word-forms. Their meaning seems to be different from, and more heterogeneous than, that of full word-forms. The difference between the two classes of word-forms comes out immediately in relation to some of the theories of meaning men­tioned in Chapter 1. For example, it might seem reasonable enough to say that the meaning of 'dog' is some kind of concept or behavioural response, which can be described or explicated without taking into account the phrases and sentences of English in which 'dog' can occur. But it hardly makes sense to discuss the meaning of the, of, and, to and if in such terms. Nor does it seem reasonable to say that their meaning, however we describe or explicate it, is independent of their grammatical function. This difference between full forms and empty forms is consistent with the fact that (as was mentioned above) the major parts of speech — especially nouns and verbs - are traditionally defined, either wholly or mainly, in terms of their meaning and indepen­dently of one another, whereas the minor parts of speech — the definite and indefinite articles, prepositions, conjunctions, etc. - are always defined in terms of their grammatical function and in relation to their potential for combining with one or

2.4 Full and empty word-forms 69

other of the major parts of speech or with such higher-level units as phrases and clauses.

The grammatical distinction between full word-forms and empty word-forms that I have explained informally and non-technically in this section is, in fact, the product of several more technical distinctions, for which readers may consult the text books referred to in the Bibliography.

Since we are not concerned here with grammatical theory for its own sake, we shall not go into the details. What is really at issue, as far as we are concerned, is the distinction between the grammar of a language and its vocabulary, or lexicon, and the distinction between grammatical and lexical meaning, which depends upon it. This is a topic that will be taken up in the following section.

There is one point that can be usefully made, however, before we proceed, on the basis of the distinction drawn in this section between full word-forms and empty word-forms. This has to do with one of the questions raised in section 1.6: which is more basic than, or logically prior to, the other, the meaning of words or the meaning of sentences? One argument for the logical priority of sentence-meaning over word-meaning, which is often presented by advocates of truth-conditional semantics, runs as follows.

(i) The meaning associated with such words as if, to and and in English cannot be defined otherwise than in terms of the contribution that they make to the meaning, of the larger units - phrases, clauses and sentences - in which they occur. The meaning of such words at least is logically sec­ondary to (i.e., dependent upon) the meaning of the sen­tences in which they occur.

(ii) But the meaning of a sentence is the product of the meaning of the words of which it is composed. So, all words, both empty and full, can (and must) be brought within the scope of the general principle that the meaning of a form is the contribution it makes to the meaning of the sentences in which it occurs.

 

 

70 I Vords as meaningful units

(iii) It is methodologically preferable to have a single notion of word-meaning applicable to all words.

(iv) If the meaning of words such as if, to and and, whose mean­ing is defined as the contribution that they make to the meaning of the sentences in which they occur, is logically secondary to sentence-meaning, the meaning of all words is logically secondary to sentence-meaning, for the mean­ing of all words can be (and by methodological decision is) defined as the contribution that they make to the meaning of the sentences in which they occur.

Now, it may or may not be the case that sentence-meaning is logically prior to, or more basic than, what is here being referred to as word-meaning. But the argument that is commonly pre­sented to support this conclusion is fallacious. It rests upon the spurious methodological principle that so-called word-meaning is homogeneous: that the meaning associated with empty word-forms such as if, to and and is in all relevant respects comparable with that of full word-forms. It also trades upon the fact that the term 'word' denotes both forms and expressions and that some forms are, as it were, more fully words than others. Fullness and emptiness, in the sense in which I have been using these terms in the present section, are, in any case, a matter of degree. The emptiest of word-forms, such as if, the and and in English, are neither expressions nor forms of expressions: as we have, seen, they are semantically and, to a certain extent grammati­cally, comparable with the morphologically bound prefixes and suffixes of inflected word-forms. To call them 'words' and then to make generalizations about word-meaning on the basis of this classification merely confuses the issue.

Confusion is further confounded by what is arguably an equi­vocal use of the term 'word-meaning'. As we shall see in the fol­lowing section, 'word-meaning' does not necessarily mean the same as 'lexical meaning'. The meaning of full word-forms com­bines both lexical and grammatical meaning. Empty word-forms may not have any lexical meaning at all; and this is what is implied by saying that they are semantically empty. It, may also be mentioned here that, as we shall see later, much of the

2.5 Lexical meaning and grammatical meaning 71

discussion of the logical priority of sentence-meaning over word-meaning that is to be found in otherwise reliable and authorita­tive works on linguistic semantics, traditional and modern, is further confused by the failure to draw the distinction between sentences and utterances. For example, it is often asserted that sentences, not words, are from the outset - in the period of language-acquisition as also in adulthood - the basic units of communication. This^:assertion must be challenged. Utterances, not sentences (in the relevant sense), are the units by means of which speakers and hearers - interlocutors - communicate with one another. Some of these utterances, being grammatically complete and well-formed, are traditionally called sentences, in what, as we shall see in Part 3, is a secondary and derivative sense of 'sentence'. Increasingly complex utterances are pro­duced by children as they pass through the several distinguish­able stages of language-acquisition; but it is a long time before any of the child's utterances can reasonably be described as sentences (in what is, in any case, an irrelevant sense of the ambiguous term'sentence').

It is lexical meaning that we are discussing in Part 2. Gram­matical meaning, not all of which can be assigned to word-forms, is largely a matter of sentence-meaning, and will there­fore be dealt with in Part 3.

 

2.5 LEXICAL MEANING AND GRAMMATICAL

MEANING

 

As was noted in the preceding section, what were there referred to as full word-forms are forms of the major parts of speech, such as nouns, verbs and adjectives. Empty word-forms, in con­trast, in English (and in other languages which in this respect are typologically similar to English) belong to a wide variety of smaller form-classes, which are defined, traditionally, in terms of their syntactic function, rather than semantically.

It is for this reason that empty word-forms are traditionally described by logicians, not as independent terms or categories, but as syncategorematic: i.e., as forms whose meaning and logical function derives from the way in which they combine

 

 

72 Words as meaningful units

with syn- the independently defined major categories. I have deliberately introduced the traditional term 'category' here (together with its less familiar derivative 'syncategorematic') because in later chapters I shall be appealing frequently to an updated version of the traditional notion of categorial mean­ing. ( The term 'categorial' beat's the same sense here as it does in the phrase 'categorial ambivalence', which Was employed in the preceding chapter.) As we shall see, categorial meaning is one part of grammatical meaning: it is that part of the meaning of lexemes (and other expressions) which derives from their being members of one category rather than another (nouns rather than verbs, verbs rather than adjectives, and so on).

The distinction between full word-forms and empty word-forms has served its purpose. I now want to introduce the distinc­tion between the grammar of a language and its vocabulary, or lexicon. Grammar and lexicon are complementary; every grammar presupposes a lexicon, and every lexicon presupposes a grammar.

The grammar of a language is traditionally regarded as a sys­tem of rules which determines how words are put together to form (grammatically well-formed) phrases, how phrases are put together to form ('grammatically well-formed) clauses, and how clauses are put together to form (grammatically well-formed) sentences. Grammatically ill-formed combinations of words, phrases and clauses - i.e., combinations which break the rules of the grammar - are traditionally described as ungrammatical. One of the major issues that has divided twenti­eth-century theorists in their discussion of the relation between semantics and grammar is the degree to which grammatical-ity 'grammatical well-formedness) is determined by meaning-fulness. This issue will be addressed in Chapter 5.

Modern linguistic theory has produced a large set of more or less traditional, alternative approaches to the grammatical analysis of natural languages, which differ from one another in various ways. Some of these are morpheme-based (rather than word-based), in that they take the morpheme to be the basic unit of grammatical analysis (for all languages). Some recognize no distinction between clauses and sentences (and use

2.5 Lexical meaning and grammatical meaning 73

the term 'sentence' for both). Some respect the traditional bipar­tite analysis of all clauses into a subject and a predicate; others do not, or, if they do, make this a matter of secondary, rather than primary, determination. This list of differences between rival approaches could be extended almost indefinitely. The dif­ferences are by no means unimportant. But.most of them are irre­levant to the issues that will confront us in this book. Such of them as are both important and relevant will be identified as we proceed.

The lexicon may be thought of as the theoretical counterpart of a dictionary, and it is frequently so described. Looked at from a psychological point of view, the lexicon is the set (or net­work) of all the lexemes in a language, stored in the brains of competent speakers, with all the linguistic information for each lexeme that is required for the production and interpretation of the sentences of the language. Although the so-called mental lex­icon has been intensively studied in recent years from a psycholo­gical (and neuropsychological) point of view, relatively little is known so far about the way in which it is stored in the brain or about the way it is accessed in the use of a language. Relatively little is known, similarly, about the mental grammar that all speakers of a language, presumably, also carry around with them in their heads. In particular, it is not known whether there is a clear-cut psychological distinction to be drawn between grammar and lexicon.

Linguists have so far found it impossible to draw any such dis­tinction sharply in the description of particular languages. And this is one reason for the controversy and lack of consensus that currently exists among linguists as to the way in which grammar and lexicon should be integrated in the systematic description of languages. This is one of the controversies that we do not need to get involved with in a book of this kind. For simplicity of exposition, I will adopt a deliberately conservative view of the relation between grammar and lexicon: the view that is reflected in standard textbooks of linguistics and in conventional dictionaries of English and other languages. Adjustments can easily be made by those readers who are familiar with recent grammatical theory (which, in this and other respects, has in

 

 

74 Words as meaningful units

any case not completely superseded traditional grammar and can still profitably draw upon it for many of the concepts that it seeks to formalize and explicate).

Although we are not concerned with grammatical theory as such in this book, we are concerned with the way in which mean­ing is encoded in the grammatical (i.e., the syntactic and the morphological) structure of languages. It was in that connexion that, in the preceding section, I introduced the distinction between what I called full word-forms and empty word-forms. Some, though not all, empty word-forms, in English and other typologically similar languages, will have a purely grammatical meaning (if they have any meaning at all). All full word-forms, on the other hand, will have both a lexical and a grammatical, and more particularly a categorial, meaning. For example, child and children, being forms of the same lexeme ('child') have the same lexical meaning (which I am symbolizing, notation-ally, as "child"). In so far as the lexeme has certain semantically relevant grammatical properties (it is a noun of particular kind), the two word-forms also share some part of their categor­ial meaning. But they differ, of course, grammatically (more precisely, morphosyntactically) in that the one is a singular and the other a plural noun-form. The difference between singular and plural (in those languages in which it is grammaticalized) is another part of the categorial component of grammatical meaning. And it is of course accounted for traditionally, both grammatically and semantically, in terms of what may be thought of as the secondary grammatical category of number. Other such semantically relevant secondary grammatical cat­egories (not all of which are to be found in all languages) are tense, mood, aspect, gender and person. Reference will be made to some of these categories in later chapters.