И. В. Арнольд
 Скачать 1.82 Mb.
|
|
§ 3.4 THE SEMANTIC STRUCTURE OF POLYSEMANTIC WORDS Polysemy is characteristic of most words in many languages, however different they may be. But it is more characteristic of the English vo- 1  There is a vast literature on the problems of denotation, connotation and implication that can be recommended as background reading. These are works by E.S. Aznaurova, M.V. Nikitin, I.V. Arnold, I.P. Sternin, V.I. Shakhovsky and others. The references are given in full at the end of the book.50 cabulary as compared with Russian, due to the monosyllabic character of English and the predominance of root words. The greater the relative frequency of the word, the greater the number of variants that constitute its semantic structure, i.e. the more polysemantic it is. This regularity is of course a statistical, not a rigid one.1 Word counts show that the total number of meanings separately registered in NED for the first thousand of the most frequent English words is almost 25,000, i.e. the average number of meanings for each of these most frequent words is 25. Consider some of the variants of a very frequent, and consequently polysemantic word run. We define the main variant as 4to go by moving the legs quickly' as in: Tired as I was, I began to run frantically home. The lexical meaning does not change in the forms ran or running. The basic meaning may be extended to inanimate things: / caught the bus that runs between C and B; or the word run may be used figuratively: // makes ike blood run cold. Both the components lon foot' and 'quickly' are suppressed in these two last examples, as well as in The car runs on petrol. The idea of motion remains but it is reduced to 'operate or function'. The difference of meaning is reflected in the difference of syntactic valency. It is impossible to use this variant about humans and say: *We humans run on food. The active-passive transformation is possible when the meaning implies 'management': The Co-op runs this self-service shop — This self-service shop is run by the Co-op, but */ was run by home is obviously nonsense. The component 'speed' is important in the following: Then though we cannot make our sun Stand still, yet we will make him run (Marvel]). There are other variants of run where there is no implication of speed or 'on foot' or motion of any kind but the seme of direction is retained: On the other side of the stream the bank ran up steeply. *The bank ran without the implication of direction is meaningless. There are also other variants of the verb run, they all have something in common with some of the others. To sum up: though there is no single semantic component common to all the lexico-semantic variants of the verb run, every variant has something in common with at least one of the others. Every meaning in language and every difference in meaning is signalled either by the form of the word itself or by context, i.e. its syntag-matic relations depending on the position in the spoken chain. The unity of the two facets of a linguistic sign — its form and its content in the case of a polysemantic word — is kept in its lexico-grammatical variant. No universally accepted criteria for differentiating these variants within one polysemantic word can so far be offered, alt hough the problem has lately attracted a great deal of attention. The main points can be 1  A special formula known as "Zipf's Law" has been worked out to express the correlation between frequency, word length and polysemy.51 summed up as follows: lexico-gramniatical variants of a word are its variants characterized by paradigmatic or morphological peculiarities, different valency, different syntactic functions; very often they belong to different lexico-grammatical groups of the same part of speech. Thus run is intransitive in / ran home, but transitive in / run this office. Some of the variants demand an object naming some vehicle, or some adverbials of direction, and so on. All the lexical and lexico-grammatical variants of a word taken together form its semantic structure or semantic paradigm. Thus, in the semantic structure of the word youth three lexico-grammatical variants may be distinguished: the first is an abstract uncountable noun, as in the frietids of one's youthKihe second is a countable personal noun 'a young man' (plural youths) that can be substituted by the pronoun he in the singular and they in the plural; the third is a collective noun 'young men and women' having onh one form, that of the singular, substituted by the pronoun they. Within the first lexico-grammatical variant two shades of meaning can be distinguished with two different referents, one denoting the state of being young, and the other the time of being young. These shades of meaning are recognized due to the lexical peculiarities of distribution and sometimes are blended together as in to feel that one's youth has gone, where both the time and the state can be meant. These variants form a structured set because they are expressed by the same sound complex and are interrelated in meaning as they all contain the semantic component 'young' and can be explained by means of one another. No general or complete scheme of types of lexical meaning as elements of a word's semantic structure has so far been accepted by linguists. Linguistic literature abounds in various terms reflecting various points of view. The following terms may be found with different authors: the meaning is direct when it nominates the referent without the help of a context, in isolation, i.e. in one word sentences. The meaning is figurative when the object is named and at the same time characterized through its similarity with another object. Note the word characterized: it is meant to point out that when used figuratively a word, while naming an object simultaneously describes it. Other oppositions are concrete:: abstract; main/ primary::secondary; central::peripheric; nar-r o w : : e x t e n d e d; general::special/particular, and so on. One readily sees that in each of these the basis of classification is different, although there is one point they have in common. In each case the comparison takes place within the semantic structure of one word. They are characterized one against the other. Take, for example, the noun screen. We find it in its direct meaning when it names a movable piece of furniture used to hide something or protect somebody, as in the case of fire-screen placed in front of a fireplace. The meaning is figurative when the word is applied to anything which protects by hiding, as in smoke screen. We define this meaning as figurative comparing it to the first that we called direct. Again, when by a screen the speaker means 'a silver-coloured sheet on which 52 pictures are shown', this meaning in comparison with the main/primary will be secondary. When the same word is used attributively in such combinations as screen actor, screen star, screen version, etc., it comes to mean 'pertaining to the cinema' and is abstract in comparison with the first meaning which is concrete. The main meaning is that which possesses the highest frequency at the present stage of vocabulary development. All these terms reflect relationships existing between different meanings of a word at the same period, so the classification may be called synchronic and paradigmatic, although the terms used are borrowed from historical lexicology and stylistics.1 If the variants are classified not only by comparing them inside the semantic structure of the word but according to the style and sphere of language in which they may occur, if they have stylistic connotations, the classification is stylistical. All the words are classified into stylistically neutral and stylistically coloured. The latter may be classified into bookish and colloquial, bookish styles in their turn may be (a) general, (b) poetical, (c) scientific or learned, while colloquial st>les are subdivided into (a) literary colloquial, (b) familiar colloquia 1, (c) slang. If we are primarily interested in the historical perspective, the meanings will be classified according to their genetic characteristic and their growing or diminishing role in the language. In this way the following terms are used: etymological, i.e. the earliest known meaning; archaic, i.e. the meaning superseded at present by a newer one but still remaining in certain collocations; obsolete, gone out of use; present-day meaning, which is the one most frequent in the present-day language and the original meaning serving as basis for the derived ones. It is very important to pay attention to the fact that one and the same meaning can at once belong, in accordance with different points, to different groups. These features of meaning ma\ therefore serve as distinctive features describing each meaning in its relationship to the others. Diachronic and synchronic ties are thus closely interconnected as the new meanings are understood thanks to their motivation by the older meanings. Hornby's dictionary, for instance, distinguishes in the word witness four different variants, which may be described as follows. witness1'evidence, testimony' — a direct, abstract, primary meaning witness2'a person who has first-hand knowledge cf an event and is able to describe it' —ja metonymical, concrete, secondary meaning witness^ 'a person who gives evidence under oath in a law court* — a metonymical, concrete, secondary meaning specialized from witness^ witness^ 'a person who puts his signature to a document by the side of that of the chief person who signs it' — a metonymical, concrete, secondary meaning specialized from wil/iess2 1 Some authors call relations within one word — epidigmatic. See p. 41. 53 Polysemy is a phenomenon of language not of speech. The sum total of many contexts in which the word is observed to occur permits the lexicographers to record cases of identical meaning and cases that differ in meaning. They are registered by lexicographers and found in dictionaries. A distinction has to be drawn between the lexical meaning of a word in speech, we shall call it contextual meaning, and the semantic structure of a word in language. Thus the semantic structure of the verb act comprises several variants: ‘do something’, ‘behave’, ‘take a part in a play’, ‘pretend’. If one examines this word in the following aphorism: Some men have acted courage who had it not; but no man can act wit (Halifax), one sees it in a definite context that particularises it and makes possible only one meaning ‘pretend’. This contextual meaning has a connotation of irony. The unusual grammatical meaning of transitivity (act is as a rule intransitive) and the lexical meaning of objects to this verb make a slight difference in the lexical meaning. As a rule the contextual meaning represents only one of the possible variants of the word but this one variant may render a complicated notion or emotion analyzable into several semes. In this case we deal not with the semantic structure of the word but with the semantic structure of one of its meanings. Polysemy does not interfere with the communicative function of the language because the situation and context cancel all the unwanted meanings. Sometimes, as, for instance in puns, the ambiguity is intended, the words are purposefully used so as to emphasise their different meanings. Consider the replica of lady Constance, whose son, Arthur Plantagenet is betrayed by treacherous allies: LYMOGES (Duke of Austria): Lady Constance, peace! CONSTANCE: War! war! no peace! peace is to me a war (Shakespeare). In the time of Shakespeare peace as an interjection meant ‘Silence!’ But lady Constance takes up the main meaning — the antonym of war. Geoffrey Leech uses the term reflected meaning for what is communicated through associations with another sense of the same word, that is all cases when one meaning of the word forms part of the listener’s response to another meaning. G. Leech illustrates his point by the following example. Hearing in the Church Service the expression The Holy Ghost, he found his reaction conditioned by the everyday unreligious and awesome meaning ‘the shade of a dead person supposed to visit the living’. The case where reflected meaning intrudes due to suggestivity of the expression may be also illustrated by taboo words and euphemisms connected with the physiology of sex. Consider also the following joke, based on the clash of different meanings of the word expose (‘leave unprotected’, ‘put up for show’, ‘reveal the guilt of’). E. g.: Painting is the art of protecting flat surfaces from the weather and exposing them to the critic. Or, a similar case: “Why did they hang this picture?” “Perhaps, they could not find the artist.” 54 Contextual meanings include nonce usage. Nonce words are words invented and used for a particular occasion. The study of means and ways of naming the elements of reality is called onomasiology. As worked out in some recent publications it received the name of Theory of Nomination.1 So if semasiology studies what it is the name points out, onomasiology and the theory of nomination have to show how the objects receive their names and what features are chosen to represent them. Originally the nucleus of the theory concerned names for objects, and first of all concrete nouns. Later on a discussion began, whether actions, properties, emotions and so on should be included as well. The question was answered affirmatively as there is no substantial difference in the reflection in our mind of things and their properties or different events. Everything that can be named or expressed verbally is considered in the theory of nomination. Vocabulary constitutes the central problem but syntax, morphology and phonology also have their share. The theory of nomination takes into account that the same referent may receive various names according to the information required at the moment by the process of communication, e. g. Walter Scott and the author of Waverley (to use an example known to many generations of linguists). According to the theory of nomination every name has its primary function for which it was created (primary or direct nomination), and an indirect or secondary function corresponding to all types of figurative, extended or special meanings (see p. 53). The aspect of theory of nomination that has no counterpart in semasiology is the study of repeated nomination in the same text, as, for instance, when Ophelia is called by various characters of the tragedy: fair Ophelia, sweet maid, dear maid, nymph, kind sister, rose of May, poor Ophelia, lady, sweet lady, pretty lady, and so on. To sum up this discussion of the semantic structure of a word, we return to its definition as a structured set of interrelated lexical variants with different denotational and sometimes also connotational meanings. These variants belong to the same set because they are expressed by the same combination of morphemes, although in different contextual conditions. The elements are interrelated due to the existence of some common semantic component. In other words, the word’s semantic structure is an organised whole comprised by recurrent meanings and shades of meaning that a particular sound complex can assume in different contexts, together with emotional, stylistic and other connotations, if any. Every meaning is thus characterised according to the function, significative or pragmatic effect that it has to fulfil as denotative and connotative meaning referring the word to the extra-linguistic reality and to the speaker, and also with respect to other meanings with which it is contrasted. The hierarchy of lexico-grammatical variants and shades of meaning within the semantic structure of a word is studied with the help of formulas establishing semantic distance between them developed by N. A. Shehtman and other authors. _____________________ 1 The problem was studied by W. Humboldt (1767-1835) who called the feature chosen as the basis of nomination— the inner form of the word. 55 § 3.5 CONTEXTUAL ANALYSIS The contextual method of linguistic research holds its own alongside statistical, structural and other developments. Like structural methods and procedures, it is based on the assumption that difference in meaning of linguistic units is always indicated by a difference in environment. Unlike structural distributional procedures (see §5.2, 5.3) it is not formalised. In some respects, nevertheless, it is more rigorous than the structural procedures, because it strictly limits its observations and conclusions to an impressive corpus of actually recorded material. No changes, whether controlled or not, are permitted in linguistic data observed, no conclusions are made unless there is a sufficient number of examples to support their validity. The size of a representative sample is determined not so much by calculation though, but rather by custom. Words are observed in real texts, not on the basis of dictionaries. The importance of the approach cannot be overestimated; in fact, as E. Nida puts it, “it is from linguistic contexts that the meanings of a high proportion of lexical units in active or passive vocabularies are learned."1 The notion of context has several interpretations. According to N. N. Amosova context is a combination of an indicator or indicating minimum and the dependant, that is the word, the meaning of which is to be rendered in a given utterance. The results until recently were, however more like a large collection of neatly organised examples, supplemented with comments. A theoretical approach to this aspect of linguistics will be found in the works by G. V. Kolshansky. Contextual analysis concentrated its attention on determining the minimal stretch of speech and the conditions necessary and sufficient to reveal in which of its individual meanings the word in question is used. In studying this interaction of the polysemantic word with the syntactic configuration and lexical environment contextual analysis is more concerned with specific features of every particular language than with language universals. Roughly, context may be subdivided into lexical, syntactical and mixed. Lexical context, for instance, determines the meaning of the word black in the following examples. Black denotes colour when used with the key-word naming some material or thing, e. g. black velvet, black gloves. When used with key-words denoting feeling or thought, it means ‘sad’, ‘dismal’, e. g. black thoughts, black despair. With nouns denoting time, the meaning is ‘unhappy’, ‘full of hardships’, e. g. black days, black period. If, on the other hand, the indicative power belongs to the syntactic pattern and not to the words which make it up, the context is called syntactic. E. g. make means ‘to cause’ when followed by a complex object: I couldn’t make him understand a word I said. _____________________ 1 Nida E. Componential Analysis of Meaning. The Hague-Paris, Mouton 1975. P. 195. 56 A purely syntactic context is rare. As a rule the indication comes from syntactic, lexical and sometimes morphological factors combined. Thus, late, when used predicatively, means ‘after the right, expected or fixed time’, as be late for school. When used attributively with words denoting periods of time, it means ‘towards the end of the period’, e. g. in late summer. Used attributively with proper personal nouns and preceded with a definite article, late means ‘recently dead’.  All lexical contexts are subdivided into lexical contexts of the first degree and lexical contexts of the second degree. In the lexical context of the first degree there is a direct syntactical connection between the indicator and the dependent: He was arrested on a treason charge. In lexical context of the second degree there is no direct syntactical connection between a dependent and the indicator. E.g.: I move that Mr Last addresses the meeting (Waugh). The dependent move is not directly connected to the indicating minimum addresses the meeting. Alongside the context N. N. Amosova distinguishes speech situation, in which the necessary indication comes not from within the sentence but from some part of the text outside it. Speech situation with her may be of two types: text-situation and life-situation. In text-situation it is a preceding description, a description that follows or some word in the preceding text that help to understand the ambiguous word. E. Nida gives a slightly different classification. He distinguishes linguistic and practical context. By practical context he means the circumstances of communication: its stimuli, participants, their relation to one another and to circumstances and the response of the listeners. 3.6 COMPONENTIAL ANALYSIS A good deal of work being published by linguists at present and dealing with semantics has to do with componential analysis.1 To illustrate what is meant by this we have taken a simple example (see p. 41) used for this purpose by many linguists. Consider the following set of words: man, woman, boy, girl, bull, cow. We can arrange them as correlations of binary oppositions man : : woman = boy : : girl = bull : : cow. The meanings of words man, boy, bull on the one hand, and woman, girl and cow, on the other, have something in common. This distinctive feature we call a semantic component or seme. In this case the semantic distinctive feature is that of sex — male or female. Another possible correlation is man : : boy = woman : : girl. The distinctive feature is that of age — adult or non-adult. If we compare this with a third correlation man : : bull = woman : : cow, we obtain a third distinctive feature contrasting human and animal beings. In addition to the notation given on p. 41, the componential formula may be also shown by brackets. The meaning of man can be described as (male (adult (human being))), woman as (female (adult (human being))), girl as (female (non-adult (human being))), etc. _____________________ 1 See the works by O.K. Seliverstova, J.N. Karaulov, E. Nida, D. Bolinger and others. 57 Componential analysis is thus an attempt to describe the meaning of words in terms of a universal inventory of semantic components and their possible combinations.1 Componential approach to meaning has a long history in linguistics.2 L. Hjelmslev’s commutation test deals with similar relationships and may be illustrated by proportions from which the distinctive features d1, d2, d3 are obtained by means of the following procedure: As the first relationship is that of male to female, the second, of young to adult, and the third, human to animal, the meaning ‘boy’ may be characterised with respect to the distinctive features d1, d2, d3 as containing the semantic elements ‘male’, ‘young’, and ‘human’. The existence of correlated oppositions proves that these elements are recognised by the vocabulary. In criticising this approach, the English linguist Prof. W. Haas3 argues that the commutation test looks very plausible if one has carefully selected examples from words entering into clear-cut semantic groups, such as terms of kinship or words denoting colours. It is less satisfactory in other cases, as there is no linguistic framework by which the semantic contrasts can be limited. The commutation test, however, borrows its restrictions from philosophy. A form of componential analysis describing semantic components in terms of categories represented as a hierarchic structure so that each subsequent category is a sub-category of the previous one is described by R. S. Ginzburg. She follows the theory of the American linguists J. Katz and J. Fodor involving the analysis of dictionary meanings into semantic markers and distinguishers but redefines it in a clear-cut way. The markers refer to features which the word has in common with other lexical items, whereas a distinguishes as the term implies, differentiates it from all other words. We borrow from R. S. Ginzburg her analysis of the word spinster. It runs as follows: spinster — noun, count noun, human, adult, female, who has never married. Parts of speech are the most inclusive categories pointing to major classes. So we shall call this component class seme (a term used by French semasiologists). As the grammatical function is predominant when we classify a word as a count noun it seems more logical to take this feature as a subdivision of a class seme. _____________________ 1 Note the possibility of different graphical representation. 2 Componential analysis proper originates with the work of F.G. Lounsbury and W.H. Goodenough on kinship terms. 3 Prof. W. Haas (of Manchester University) delivered a series of lectures on the theory of meaning at the Pedagogical Institutes of Moscow and Leningrad in 1965. 58 It may, on the other hand, be taken as a marker because it represents a sub-class within nouns, marks all nouns that can be counted, and differentiates them from all uncountable nouns. Human is the next marker which refers the word spinster to a sub-category of nouns denoting human beings (man, woman, etc. vs table, flower, etc.). Adult is another marker pointing at a specific subdivision of living beings into adult and not grown-up (man, woman vs boy, girl). Female is also a marker (woman, widow vs man, widower), it represents a whole class of adult human females. ‘Who has never married’ — is not a marker but a distinguisher, it differentiates the word spinster from other words which have other features in common (spinster vs widow, bride, etc.). The analysis shows that the dimensions of meaning may be regarded as semantic oppositions because the word’s meaning is reduced to its contrastive elements. The segmentation is continued as far as we can have markers needed for a group of words, and stops when a unique feature is reached. A very close resemblance to componential analysis is the method of logical definition by dividing a genus into species and species into subspecies indispensable to dictionary definitions. It is therefore but natural that lexicographic definitions lend themselves as suitable material for the analysis of lexical groups in terms of a finite set of semantic components. Consider the following definitions given in Hornby’s dictionary: cow — a full grown female of any animal of the ox family calf — the young of the cow The first definition contains all the elements we have previously obtained from proportional oppositions. The second is incomplete but we can substitute the missing elements from the previous definition. We can, consequently, agree with J. N. Karaulov and regard as semantic components (or semes) the notional words of the right hand side of a dictionary entry. It is possible to describe parts of the vocabulary by formalising these definitions and reducing them to some standard form according to a set of rules. The explanatory transformations thus obtained constitute an intersection of transformational and componential analysis. The result of this procedure applied to collective personal nouns may be illustrated by the following.  e. g. team → a group of people acting together in a game, piece of work, etc. Procedures briefly outlined above proved to be very efficient for certain problems and find an ever-widening application, providing us with a deeper insight into some aspects of language.1 _____________________ 1 For further detail see: Арнольд И.В. Семантическая структура слова в современном английском языке и методика ее исследования. Л., 1966. 59 Chapter 4 SEMANTIC CHANGE |