THE CRITERIA OF COMPOUNDS

As English compounds consist of free forms, it is difficult to distinguish them from phrases. The combination top dog ‘a person occupying foremost place’, for instance, though formally broken up, is neither more nor less analysable semantically than the combination underdog ‘a person who has the worst of an encounter’, and yet we count the first (top dog) as a phrase and the second (underdog) as a word. How far is this justified? In reality the problem is even more complex than this isolated example suggests. Separating compounds from phrases and also from derivatives is no easy task, and scholars are not agreed upon the question of relevant criteria. The following is a brief review of various solutions and various combinations of criteria that have been offered.

The problem is naturally reducible to the problem of defining word boundaries in the language. It seems appropriate to quote E. Nida who writes that “the criteria for determining the word-units in a language are of three types: (1) phonological, (2) morphological, (3) syntactic. No one type of criteria is normally sufficient for establishing the word-unit. Rather the combination of two or three types is essential."¹

E. Nida does not mention the graphic criterion of solid or hyphenated spelling. This underestimation of written language seems to be a mistake. For the present-day literary language, the written form is as important as the oral. If we accept the definition of a written word as the part of the text from blank to blank, we shall have to accept the graphic criterion as a logical consequence. It may be argued, however, that there is no consistency in English spelling in this respect. With different dictionaries and different authors and sometimes even with the same author the spelling varies, so that the same unit may exist in a solid spelling: headmaster, loudspeaker, with a hyphen: head-master, loud-speaker and with a break between the components: head master, loud speaker. Compare also: airline, air-line, air line’, matchbox, matchbox, match box’, break-up, breakup. Moreover, compounds that appear to be constructed on the same pattern and have similar semantic relations between the constituents may be spelt differently: textbook, phrase-book and reference book. Yet if we take into consideration the comparative frequency of solid or hyphenated spelling of the combinations in question, the criterion is fairly reliable. These three types of spelling need not indicate different degrees of semantic fusion. Sometimes hyphenation may serve aesthetic purposes, helping to avoid words that will look too long, or purposes of convenience, making syntactic components clearer to the eye: peace-loving nations, old-fashioned ideas.

¹ Nida E. Morphology. P. 147; Quirk R. et al. A Grammar of Contemporary English. P. 1019.

This lack of uniformity in spelling is the chief reason why many authors consider this criterion insufficient. Some combine it with the phonic criterion of stress. There is a marked tendency in English to give compounds a heavy stress on the first element. Many scholars consider this unity of stress to be of primary importance. Thus L. Bloomfield writes: “Wherever we hear lesser or least stress upon a word which would always show a high stress in a phrase, we describe it as a compound member: ice-cream ['ajs-krijm] is a compound but ice cream ['ajs'krijm] is a phrase, although there is no denotative difference in meaning."¹

It is true that all compound nouns, with very few exceptions, are stressed on this pattern. Cf. ‘blackboard : : ‘blackboard’, ‘blackbird : : ‘black'bird; ‘bluebottle : : ‘blue'bottle. In all these cases the determinant has a heavy stress, the determinatum has the middle stress. The only exception as far as compound nouns are concerned is found in nouns whose first elements are all- and self-, e. g. ‘All-'Fools-Day, ‘self-con'trol. These show double even stress.

The rule does not hold with adjectives. Compound adjectives are double stressed like ‘gray-'green, ‘easy-'going, ‘new-'born. Only compound adjectives expressing emphatic comparison are heavily stressed on the first element: ‘snow-white, ‘dog-cheap.

Moreover, stress can be of no help in solving this problem because word-stress may depend upon phrasal stress or upon the syntactic function of the compound. Thus, light-headed and similar adjectives have a single stress when used attributively, in other cases the stress is even. Very often the stress is structurally determined by opposition to other combinations with an identical second element, e. g. ‘dining table : : ‘writing table. The forestress here is due to an implicit contrast that aims at distinguishing the given combination from all the other similar cases in the same series, as in ‘passenger train, ‘ freight train, ex'press train. Notwithstanding the unity stress, these are not words but phrases.

Besides, the stress may be phonological and help to differentiate the meaning of compounds:

'overwork ‘extra work'

'over'work ‘hard work injuring one’s health'

'bookcase ‘a piece of furniture with shelves for books'

^'book'case ‘a paper cover for books'

'man'kind ‘the human race'

'mankind ‘men’ (contrasted with women)

'toy,factory ‘factory that produces toys'

'toy'factory ‘factory that is a toy’.

It thus follows that phonological criterion holds for certain types of words only.²

¹ Bloomfield L. Language. P. 228. Transcription is given] as L. Bloomfield has it.

² For details see: Quirk R. et al. A Grammar of Contemporary English. Appendix 2, p.p. 1039-1042.

8 И. B. Apнольд 113

H. Paul, O. Jespersen, E. Kruisinga¹ and many others, each in his own way, advocate the semantic criterion, and define a compound as a combination forming a unit expressing a single idea which is not identical in meaning to the sum of the meanings of its components in a free phrase. From this point of view dirty work with the figurative meaning ‘dishonorable proceedings’ is a compound, while clean work or dry work are phrases. Сf. fusspot, slow-coach. The insufficiency of this criterion will be readily understood if one realises how difficult it is to decide whether the combination in question expresses a single integrated idea. Besides, between a clearly motivated compound and an idiomatic one there are a great number of intermediate cases. Finally, what is, perhaps, more important than all the rest, as the semantic features and properties of set expressions are similar to those of idiomatic compounds, we shall be forced to include all idiomatic phrases into the class of compounds. Idiomatic phrases are also susceptible to what H. Paul calls isolation, since the meaning of an idiomatic phrase cannot be inferred from the meaning of components. For instance, one must be specially explained the meaning of the expressions (to rain) cats and dogs, to pay through the nose, etc. It cannot be inferred from the meaning of the elements.

As to morphological criteria of compounds, they are manifold. Prof. A. I. Smirnitsky introduced the criterion of formal integrity.² He compares the compound shipwreck and the phrase (the) wreck of (a) ship comprising the same morphemes, and points out that although they do not differ either in meaning or reference, they stand in very different relation to the grammatical system of the language. It follows from his example that a word is characterised by structural integrity non-existent in a phrase. Unfortunately, however, in the English language the number of cases when this criterion is relevant is limited due to the scarcity of morphological means.

“A Grammar of Contemporary English” lists a considerable number of patterns in which plural number present in the correlated phrase is neutralised in a compound. Taxpayer is one who pays taxes, cigar smoker is one who smokes cigars, window-cleaner is one who cleans windows, lip-read is to read the lips. The plural of still-life (a term of painting) is still-lifes and not still lives. But such examples are few. It cannot be overemphasised that giving a mere description of some lexicological phenomenon is not enough; one must state the position of the linguistic form discussed in the system of the language, i.e. the relative importance of the type. Therefore the criterion of structural integrity is also insufficient.

The same is true as regards connective elements which ensure the integrity. The presence of such an element leaves no doubt that the combination

¹ Paul H. Prinzipien der Sprachgeschichte. 3 Aufl., Halle, 1898. S. 302; Kruisinga E. A Handbook of Present-Day English. Gröningen, 1932. Pt. II. P. 72; Jespersen O. A Modern English Grammar on Historical Principles. London, 1946. Pt. VI. P. 137.

² See: Cмирницкий А.И. Лексикология английского языка. M., 1956. С. 33.

is a compound but the number of compounds containing connective elements is relatively insignificant. These elements are few even in languages morphologically richer than English. In our case they are -s- (craftsman), -o- (Anglo-Saxon), -i- (handiwork.)

Diachronically speaking, the type craftsman is due either to the old Genitive (guardsman, kinsman, kinswoman, sportsman, statesman, tradesman, tradeswoman, tradesfolk, tradespeople) or to the plural form.

The Genitive group is kept intact in the name of the butterfly death’s head and also in some metaphorical plant names: lion’s snout, bear’s ear, heart’s ease, etc.

The plural form as the origin of the connective -s- is rarer: beeswax, woodsman, salesman, saleswoman. This type should be distinguished from clothes-basket, goods-train or savings-bank, where the singular form of the word does not occur in the same meaning.

It has already been pointed out that the additive (copulative) compounds of the type Anglo-Saxon are rare, except in special political or technical literature.

Sometimes it is the structural formula of the combination that shows it to be a word and not a phrase. E. g. starlit cannot be a phrase because its second element is the stem of a participle and a participle cannot be syntactically modified by a noun. Besides the meaning of the first element implies plurality which should have been expressed in a phrase. Thus, the word starlit is equivalent to the phrase lit by stars.

It should be noted that lit sounds somewhat, if a very little, obsolete: the form lighted is more frequent in present-day English. This survival of obsolete forms in fixed contexts or under conditions of fixed distribution occurs both in phraseology and composition.

To some authors the syntactical criterion based on comparing the compound and the phrase comprising the same morphemes seems to ,be the most promising. L. Bloomfield points out that “the word black in the phrase black birds can be modified by very (very black birds) but not so the compound-member black in blackbirds."¹ This argument, however, does not permit the distinguishing of compounds from set expressions any more than in the case of the semantic criterion: the first element of black market or black list (of persons under suspicion) cannot be modified by very either.²

This objection holds true for the argument of indivisibility advanced by B. Bloch and G. Trager who point out that we cannot insert any word between the elements of the compound blackbird.³ Thesame example black market serves H. Marchand to prove the insufficiency of this criterion.⁴ Black market is indivisible and yet the stress pattern shows it is a phrase.

¹ Bloomfield L. Language. P. 232.

² Prof. R. Lord in his letter to the author expressed the opinion that black market and black list could be modified by very in order to produce an ironically humorous effect, although admittedly this kind of thing would not occur in normal speech. The effect of the deviation therefore proves the existence of the norm.

³ Bloch B. and Trager G. Outline of Linguistic Analysis. P. 66.

⁴ Marchand H. The Categories and Types .... P. 14.

Some transformational procedures that have been offered may also prove helpful. The gist of these is as follows. A phrase like a stone wall can be transformed into the phrase a wall of stone, whereas a toothpick cannot be replaced by a pick for teeth. It is true that this impossibility of transformation proves the structural integrity of the word as compared with the phrase, yet the procedure works only for idiomatic compounds, whereas those that are distinctly motivated permit the transformation readily enough:

a toothpick ↔ a pick for teeth tooth-powder → powder for teeth a tooth-brush → a brush for teeth

In most cases, especially if the transformation is done within the frame of context, this test holds good and the transformation, even if it is permissible, brings about a change of meaning. For instance, ...the wall-papers and the upholstery recalled ... the refinements of another epoch (Huxley) cannot be transformed without ambiguity into the papers on the wall and the upholstery recalled the refinements of another epoch.

That is why we shall repeat with E. Nida that no one type of criteria is normally sufficient for establishing whether the unit is a compound or a phrase, and for ensuring isolation of word from phrase. In the majority of cases we have to depend on the combination of two or more types of criteria (phonological, morphological, syntactic or graphical). But even then the ground is not very safe and the path of investigation inevitably leads us to the intricate labyrinth of “the stone wall problem” that has received so much attention in linguistic literature. (See p. 118.)