Considering this peculiarity of lexico-grammatical homonyms we may subdivide them into two groups: A. identical in sound-form but different in their grammatical and lexical meanings (s eal1 n — seal3 v), and B. identical in sound-form but different in their grammatical meanings and partly different in their lexical meaning, i.e. partly different in their semantic structure (seal3 n — seal3 v; paper n — (to) paper v). Thus the definition of homonyms as words possessing identical sound-form but different semantic structure seems to be more exact as it allows of a better understanding of complex cases of homonymy, e.g. seal1 n — seal2 n; seal3 v — seal4 v which can be analysed into homonymic pairs, e.g. seal1 n — seal 2 n lexical homonyms; seal1 n — seal3 v — lexico-
grammatical homonyms, subgroup A; seal2 n — seal3 v — lexico-grammatical homonyms, subgroup B.
|§ 35. Graphic and Sound-Form of Homonyms
In the discussion of the problem of homonymy we proceeded from the assumption that words are two-facet units possessing both sound-form and meaning, and we deliberately disregarded their graphic form. Some linguists, however, argue that the graphic form of words in Modern English is just as important as their sound-form and should be taken into consideration in the analysis and classification ■ of homonyms. Consequently they proceed from definition of homonyms as words identical in sound-form or spelling but different in meaning. It follows that in their classification of homonyms all the three aspects: sound-form, graphic form and meaning are taken into account. Accordingly they classify homonyms into homographs, homophones and perfect homonyms.
Homographs are words identical in spelling, but different both in their sound-form and meaning, e.g. bow n [bou] — ‘a piece of wood curved by a string and used for shooting arrows’ and bow n [bau] — ‘the bending of the head or body’; tear n [tia] — ‘a drop of water that comes from the eye’ and tear v [tea] — ‘to pull apart by force’.
Homophones are words identical in sound-form but different both in spelling and in meaning, e.g. sea n and see v; son n and sun n.
Perfect homonyms are words identical both in spelling and in sound-form but different in meaning, e.g. case1 n — ’something that has happened’ and case2 n — ‘a box, a container’.
|§ 36. Sources of Homonymy
The description of various types of homonyms in Modern English would be incomplete if we did not give a brief outline of the diachronic processes that account for their appearance.
The two main sources of homonymy are: 1) diverging meaning development of a polysemantic word, and 2) converging sound development of two or more different words. The process of diverging meaning development can be observed when different meanings of the same word move so far away from each other that they come to be regarded as two separate units. This happened, for example, in the case of Modern English flower and flour which originally were one word (ME. flour, cf. OFr. flour, flor, L. flos — florem) meaning ‘the flower’ and ‘the finest part of wheat’. The difference in spelling underlines the fact that from the synchronic point of view they are two distinct words even though historically they have a common origin.
Convergent sound development is the most potent factor in the creation of homonyms. The great majority of homonyms arise as a result of converging sound development which leads to the coincidence of two or more words which were phonetically distinct at an earlier date. For example, OE. ic and OE. åàzå have become identical in pronunciation (MnE. I [ai] and eye [ai]). A number of lexico-grammatical homonyms appeared as a result of convergent sound development of the verb and the noun (cf. MnE. love — (to) love and OE. lufu — lufian).
Words borrowed from other languages may through phonetic convergence become homonymous. ON. ras and Fr. race are homonymous in Modern English (cf. race1 [ reis] — ‘running’ and race2 [ reis] — ‘a distinct ethnical stock’).
|§ 37. Polysemy and Homonymy: Etymological and Semantic Criteria
One of the most debatable problems in semasiology is the demarcation line between homonymy and polysemy, i.e. between different meanings of one word and the meanings of two homonymous words.
If homonymy is viewed diachronically then all cases of sound convergence of two or more words may be safely regarded as cases of homonymy, as, e.g., race1 and race2 can be traced back to two etymologically different words. The cases of semantic divergence, however, are more doubtful. The transition from polysemy to homonymy is a gradual process, so it is hardly possible to point out the precise stage at which divergent semantic development tears asunder all ties between the meanings and results in the appearance of two separate words. In the case of flower, flour, e.g., it is mainly the resultant divergence of graphic forms that gives us grounds to assert that the two meanings which originally made up the semantic structure of îne word are now apprehended as belonging to two different words.
Synchronically the differentiation between homonymy and polysemy is as a rule wholly based on the semantic criterion. It is usually held that if a connection between the various meanings is apprehended by the speaker, these are to be considered as making up the semantic structure of a polysemantic word, otherwise it is a case of homonymy, not polysemy.
Thus the semantic criterion implies that the difference between polysemy and homonymy is actually reduced to the differentiation between related and unrelated meanings. This traditional semantic criterion does not seem to be reliable, firstly, because various meanings of the same word and the meanings of two or more different words may be equally apprehended by the speaker as synchronically unrelated. For instance, the meaning ‘a change in the form of a noun or pronoun’ which is usually listed in dictionaries as one of the meanings of case1 seems to be synchronically just as unrelated to the meanings of this word as ’something that has happened’, or ‘a question decided in the court of law’ to the meaning of case2 — ‘abox, a container’, etc.
Secondly, in the discussion of lexico-grammatical homonymy it was pointed out that some of the meanings of homonyms arising from conversion (e.g. seal2 n — seal3 v; paper n — paper v) are related, so this criterion cannot be applied to a large group of homonymous word-forms in Modern English. This criterion proves insufficient in the synchronic analysis of a number of other borderline cases, e.g. brother — brothers — ’sons of the same parent’ and brethren — ‘fellow members of a religious society’. The meanings may be apprehended as related and then we can speak of polysemy pointing out that the difference in the morphological structure of the plural form reflects the difference of meaning. Otherwise we may regard this as a case of partial lexical homonymy.
It is sometimes argued that the difference between related and unrelated meanings may be observed in the manner in which the meanings of polysemantic words are as a rule relatable. It is observed that different meanings of one word have certain stable relationship which are not to be found ‘between the meanings of two homonymous words. A clearly perceptible connection, e.g., can be seen in all metaphoric or metonymic meanings of one word (cf., e.g., foot of the man — foot of the mountain, loud voice — loud colours, etc.,1 cf. also deep well and deep knowledge, etc.).
Such semantic relationships are commonly found in the meanings of one word and are considered to be indicative of polysemy. It is also suggested that the semantic connection may be described in terms of such features as, e.g., form and function (cf. horn of an animal and horn as an instrument), or process and result (to run — ‘move with quick steps’ and a run — act of running).
Similar relationships, however, are observed between the meanings of two partially homonymic words, e.g. to run and a run in the stocking.
Moreover in the synchronic analysis of polysemantic words we often find meanings that cannot be related in any way, as, e.g. the meanings of the word case discussed above. Thus the semantic criterion proves not only untenable in theory but also rather vague and because of this impossible in practice as in many cases it cannot be used to discriminate between several meanings of one word and the meanings of two different words.
|§ 38. Formal Criteria: Distribution and Spelling
The criterion of distribution suggested by some linguists is undoubtedly helpful, but mainly in cases of lexico-grammatical and grammatical homonymy. For example, in the homonymic pair paper «— (to) paper v the noun may be preceded by the article and followed by a verb; (to) paper can never be found in identical distribution. This formal criterion can be used to discriminate not only lexico-grammatical but also grammatical homonyms, but it often fails in cases of lexical homonymy, not differentiated by means of spelling.
Homonyms differing in graphic form, e.g. such lexical homonyms as knight — night or flower — flour, are easily perceived to be two different lexical units as any formal difference of words is felt as indicative of the existence of two separate lexical units. Conversely lexical homonyms identical both in pronunciation and spelling are often apprehended as different meanings of one word.
It is often argued that in general the context in which the words are used suffices to establish the borderline between homonymous words, e.g. the meaning of case1 in several cases of robbery can be easily differentiated from the meaning of case2 in a jewel case, a glass case. This however is true of different meanings of the same word as recorded in dictionaries, e.g. of case, as can be seen by comparing the case will be tried in the law-court and the possessive case of the noun.
1 See ‘Semasiology’, § 23, p. 31. 44
Thus, the context serves to differentiate meanings but is of little help in distinguishing between homonymy and polysemy. Consequently we have to admit that no formal means have as yet been found to differentiate between several meanings of one word and the meanings of its homonyms.
In the discussion of the problems of polysemy and homonymy we proceeded from the assumption that the word is the basic unit of language.1 Some linguists hold that the basic and elementary units at the semantic level of language are the lexico-semantic variants of the word, i.e. individual word-meanings. In that case, naturally, we can speak only of homonymy of individual lexico-semantic variants, as polysemy is by definition, at least on the synchronic plane, the coexistence of several meanings in the semantic structure of the word.
|§ 39. Summary and Conclusions
1. Homonyms are words that sound alike but have different semantic structure. The problem of homonymy is mainly the problem of differentiation between two different semantic structures of identically sounding words.
2. Homonymy of words and homonymy of individual word-forms may be regarded as full and partial homonymy. Cases of full homonymy are generally observed in words belonging to the same part of speech. Partial homonymy is usually to be found in word-forms of different parts of speech.
3. Homonymous words and word-forms may be classified by the type of meaning that serves to differentiate between identical sound-forms. Lexical homonyms differ in lexical meaning, lexico- grammatical in both lexical and grammatical meanings, whereas grammatical homonyms are those that differ in grammatical meaning only.
Lexico-grammatical homonyms are not homogeneous. Homonyms arising from conversion have some related lexical meanings in their semantic structure. Though some individual meanings may be related the whole of the semantic structure of homonyms is essentially different.
5. If the graphic form of homonyms is taken into account, they are classified on the basis of the three aspects — sound-form, graphic form and meaning — into three big groups: homographs (identical graphic form), homophones ‘ (identical sound-form) and perfect homonyms (identical sound-form and graphic form).
6. The two main sources of homonymy are: 1) diverging meaning development of a polysemantic word, and 2) convergent sound development of two or more different words. The latter is the most potent factor in the creation of homonyms.
7. The most debatable problem of homonymy is the demarcation line “between homonymy and polysemy, i.e. between different meanings of one word and the meanings of two or more phonemically different words.
1 See ‘Introduction’, § 2.
8. The criteria used in the synchronic analysis of homonymy are: 1) the semantic criterion of related or unrelated meanings; 2) the criterion of spelling; 3) the criterion of distribution.
There are cases of lexical homonymy when none of the criteria enumerated above is of any avail. In such cases the demarcation line between polysemy and homonymy is rather fluid.
9. The problem of discriminating between polysemy and homonymy in theoretical linguistics is closely connected with the problem of the basic unit at the semantic level of analysis.
WORD-MEANING IN SYNTAGMATICS AND PARADIGMATICS
It is more or less universally recognised that word-meaning can be perceived through intralinguistic relations that exist between words. This approach does not in any way deny that lexical items relate to concrete features of the real world but it is suggested that word-meaning is not comprehensible solely in terms of the referential approach.1
Intralinguistic relations of words are basically of two main types: syntagmatic and paradigmatic.
Syntagmatic relations define the meaning the word possesses when it is used in combination with other words in the flow of speech. For example, compare the meaning of the verb to get in He got a letter, He got tired, He got to London and He could not get the piano through the door.
Paradigmatic relations are those that exist between individual lexical items which make up one of the subgroups of vocabulary items, e.g. sets of synonyms, lexico-semantic groups, etc.
Paradigmatic relations define the word-meaning through its interrelation with other members of the subgroup in question. For example, the meaning of the verb to get can be fully understood only in comparison with other items of the synonymic set: get, obtain, receive, etc. Cf. He got a letter, he received a letter, he obtained a letter, etc. Comparing the sentences discussed above we may conclude that an item in a sentence can be usually substituted by one or more than one other items that have identical part-of-speech meaning and similar though not identical lexical meaning.
The difference in the type of subgroups the members of which are substitutable in the flow of speech is usually described as the difference between closed and open se,ts of lexical items. For example, any one of a number of personal pronouns may occur as the subject of a sentence and the overall sentence structure remains the same. These pronouns are strictly limited in number and therefore form a closed system in which to say he is to say not I, not you, etc. To some extent the meaning of he is defined by the other items in the system (cf., e.g., the English I, you, etc., and the Russian ÿ, òû, âû, etc.).Thesets of items in which the choice
1 See ‘Semasiology’, § 4, p. 18. 46
is limited to a finite number of alternatives as here are described as closed systems.
The members of closed systems are strictly limited in number and no addition of new items is possible.
The sets in which the number of alternatives is practically infinite as they are continually being adapted to new requirements by the addition of new lexical items are described as open systems. Closed systems are traditionally considered to be the subject matter of grammar, open systems such as lexico-semantic fields, hyponymic, synonymic sets, etc.1 are studied by lexicology.
The distinction between syntagmatic and paradigmatic relations is conventionally indicated by horizontal and vertical presentation as is shown below.
|§ 40. Polysemy and Context
From the discussion of the paradigmatic and syntagmatic relations it follows that a full understanding of the semantic structure of any lexical item can be gained only from the study of a variety of contexts in which the word is used, i.e. from the study of the intralinguistic relations of words in the flow of speech. This is of greatest importance in connection with the problem of the synchronic approach to polysemy.
It will be recalled that in analysing the semantic structure of the polysemantic word table we observed that some meanings are representative of the word in isolation, i.e. they invariably occur to us when we hear the word or see it written on paper. Other meanings come to the fore only when the word is used in certain contexts. This is true of all polysemantic words. The adjective yellow, e.g., when used in isolation is understood to denote a certain colour, whereas other meanings of this word, e.g. ‘envious’, ‘suspicious’ or ‘sensational’, ‘corrupt’, are perceived only in certain contexts, e.g. ‘a yellow look’, ‘the yellow press’, etc.
As can be seen from the examples discussed above we understand by the term context the minimal stretch of speech determining each individual meaning of the word. This is not to imply that polysemantic words have meanings only in the context. The semantic structure of the word has an objective existence as a dialectical entity which embodies
1 See ‘Semasiology’, §§ 45-50, pp. 51-61.
dialectical permanency and variability. The context individualises the meanings, brings them out. It is in this sense that we say that meaning is determined by context.
The meaning or meanings representative of the semantic structure of the word and least dependent on context are usually described as free or denominative meanings. Thus we assume that the meaning ‘a piece of furniture’ is the denominative meaning of the word table, the meaning ‘construct, produce’ is the free or denominative meaning of the verb make.
The meaning or meanings of polysemantic words observed only in certain contexts may be viewed as determined either by linguistic (or verbal) contexts or extra-linguistic (non-verbal) contexts.
The two more or less universally recognised main types of linguistic contexts which serve to determine individual meanings of words are the lexical context and the grammatical context. These types are differentiated depending on whether the lexical or the grammatical aspect is predominant in determining the meaning.
|§ 41. Lexical Context
In lexical contexts of primary importance are the groups of lexical items combined with the polysemantic word under consideration. This can be illustrated by analysing different lexical contexts in which polysemantic words are used. The adjective heavy, e.g., in isolation is understood as meaning ‘of great weight, weighty’ (heavy load, heavy table, etc.). When combined with the lexical group of words denoting natural phenomena such as wind, storm, snow, etc., it means ’striking, falling with force, abundant’ as can be seen from the contexts, e.g. heavy rain, wind, snow, storm, etc. In combination with the words industry, arms, artillery and the like, heavy has the meaning ‘the larger kind of something’ as in heavy industry, heavy artillery, etc.
The verb take in isolation has primarily the meaning ‘lay hold of with the hands, grasp, seize’, etc. When combined with the lexical group of words denoting some means of transportation (e.g. to take the tram, the bus, the train, etc.) it acquires the meaning synonymous with the meaning of the verb go.
It can be easily observed that the main factor in bringing out this or that individual meaning of the words is the lexical meaning of the words with which heavy and take are combined. This can be also proved by the fact that when we want to describe the individual meaning of a polysemantic word, we find it sufficient to use this word in combination with some members of a certain lexical group. To describe the meanings of the word handsome, for example, it is sufficient to combine it with the following words — a) man, person, b) size, reward, sum. The meanings ‘good-looking’ and ‘considerable, ample’ are adequately illustrated by the contexts.
The meanings determined by lexical contexts are sometimes referred to as lexically (or phraseologically) bound meanings which implies that such meanings are to be found only in certain lexical contexts.
Some linguists go so far as to assert that word-meaning in general can be analysed through its collocability with other words. They hold the view that if we know all the possible collocations (or word-groups) into
which a polysemantic word can enter, we know all its meanings. Thus, the meanings of the adjective heavy, for instance, may be analysed through its collocability with the words weight, safe, table; snow, wind, rain; industry, artillery, etc.
The meaning at the level of lexical contexts is sometimes described as meaning by collocation.1
|§ 42. Grammatical Context
In grammatical contexts it is the grammatical (mainly the syntactic) structure of the context that serves to determine various individual meanings of a polysemantic word. One of the meanings of the verb make, e.g. ‘to force, to enduce’, is found only in the grammatical context possessing the structure to make somebody do something or in other terms this particular meaning occurs only if the verb make is followed by a noun and the infinitive of some other verb (to make smb. laugh, go, work, etc.). Another meaning of this verb ‘to become’, ‘to turn out to be’ is observed in the contexts of a different structure, i.e. make followed by an adjective and a noun (to make a good wife, a good teacher, etc.).
Such meanings are sometimes described as grammatically (or structurally) bound meanings. Cases of the type she will make a good teacher may be referred to as syntactically bound meanings, because the syntactic function of the verb make in this particular context (a link verb, part of the predicate) is indicative of its meaning ‘to become, to turn out to be’. A different syntactic function of the verb, e.g. that of the predicate (to make machines, tables, etc.) excludes the possibility of the meaning ‘to become, turn out to be’.
In a number of contexts, however, we find that both the lexical and the grammatical aspects should be taken into consideration. The grammatical structure of the context although indicative of the difference between the meaning of the word in this structure and the meaning of the same word in a different grammatical structure may be insufficient to indicate in whiñh of its individual meanings the word in question is used. If we compare the contexts of different grammatical structures, e.g. to take+ nown and to take to+noun, we can safely assume that they represent different meanings of the verb to take, but it is only when we specify the lexical context, i.e. the lexical group with which the verb is combined in the structure to take + noun (to take coffee, tea; books, pencils; the bus, the tram) that we can say that the context determines the meaning.
It is usual in modern linguistic science to use the terms pattern or struñture to denote grammatical contexts. Patterns may be represented in conventional symbols, e.g. to take smth. as take +N. to take to smb. as take to+ N. 2 It is argued that difference in the distribution of the word is indicative of the difference in meaning. Sameness of
1 See also ‘Methods and Procedures of Lexicological Analysis’, § 4, p. 246.
2 See ‘Semasiology’, § 3, p. 1-7. Conventional symbols habitually used in distributional patterns are as follows:
N — stands for nouns or their functional equivalents, e.g. personal pronouns. V — stands for verbs except auxiliary and modal verbs (be, have, shall, etc.). A — stands for adjectives or their functional equivalents, e.g. ordinal numerals. D — stands for adverbs or their functional equivalents, e.g. at home.
distributional pattern, however, does not imply sameness of meaning. As was shown above, the same pattern to take + N may represent different meanings of the verb to take dependent mainly on the lexical group of the nouns with which it is combined.
|§ 43. Extra-Linguistic Context (Context of Situation)
Dealing with verbal contexts we consider only linguistic factors: lexical groups of words, syntactic structure of the context and so on. There are cases, however, when the meaning of the word is ultimately determined not by these linguistic factors, but by the actual speech situation in which this word is used. The meanings of the noun ring, e.g. in to give somebody a ring, or of the verb get in I've got it are determined not only by the grammatical or lexical context, but much more so by the actual speech situation.
The noun ring in such context may possess the meaning ‘a circlet of precious metal’ or ‘a call on the telephone’; the meaning of the verb to get in this linguistic context may be interpreted as ‘possess’ or ‘understand’ depending on the actual situation in which these words are used. It should be pointed out however that such cases, though possible, are not actually very numerous. The linguistic context is by far a more potent factor in determining word-meaning.