There are quite a few definitions of compounds, definitions that are widely accepted and fairly well comprehensible to ordinary readers. Zandvoort (1966: 277) writes to say that "English has a great many examples of vocables which, though felt and used as single words, are made up of two or more elements each of which may also be used as a separate word. Such vocables are called compounds". Harris (1951: 330) offers a more precise definition, having the following to say: "A compound is an English construction with two free morphemes, each with zero or more bound morphemes plus the contour". The emphasis laid on the contour seems to be crucial: viewed from the phonetic level, the elements are linked up by means of one primary stress, or they have the double-stress. Bloomfield (1933: 227) went so far as to claim for a compound to be only such a word complex which lacked the double-stress. Yet a problem arises: such complexes as breakwater, pickpocket, cupboard are, undoubtedly, regarded as compounds, as well as blackbird versus black bird, and similar; how to treat, for instance, fellow-man, though? Is this a compound, and if not, what is it then? One generally accepted way out of the problem is a further sub-classification. Jespersen (1966: 31, 47, 85) refers to old (original) compounds, recent compounds, and certain loose collocations, and Gleason (1965: 180) prefers to speak of compounds and noun phrases. It is shown, put briefly, that in original compounds only the first element is stressed (because of its being the distinctive part of the complex), e.g., goldfish, blackbird, statesman, with occasional complete phonetic obscurity in the latter element, as in cupboard, forehead, or possibly in both, e.g., shepherd. On the other hand, there are complexes composed of two elements, either receiving stress of its own, which have been called "loose collocations", e.g., stone wall, gold coin, and which may easily become what is known as "recent compounds", e.g., country town, home-made, archbishop, head-master However , the relation between phonetic contours and respective complexes is not any more straightforward. As for the stress phenomena, there are double-stressed compounds subject to the rules of the sentence rhythm, e.g., afternoon -- this afternoon -- afternoon tea; furthermore, some English compounds seem to stand very close to the category of derived words, such as Frenchman, Sunday, over-peopled, to speak nothing of certain original compounds that are treated synchronically as monophones, e.g., gospel (from the Old English 'god spell'), or which resemble compounds in spelling only, e.g., forecastle. The complexes showing a tendency towards becoming recent compounds are not easy to follow either; one can only wonder what it is that tells the native speaker to accentuate differently 'country 'town, 'country folk, 'countryman, 'country music, country 'house, (and also country 'seat) while both country 'dance and 'country dance are possible. A phonetician may be tempted to ascribe such variations to prosody, namely to rhythmical circumstances, as illustrated above by afternoon, but this is not always the case. It can also be noted that a change in stress position will sometimes bring about a change in meaning. Good examples are listed in Hockett (1966: 316-317): a 'black 'bird does not have to be a 'blackbird; a 'black 'board is not always a specially worked out 'blackboard, and although the 'White House could be seen as a 'white 'house, what it actually denotes is the U.S. President's residence. Similarly, a change in meaning can be observed in the noun phrase woman doctor: whereas the double- stressed 'woman 'doctor is any doctor who is a woman, i.e. simply 'a female doctor', 'woman doctor is an informal synonym of a 'gynaecologist', be it a man or a woman. Studying Hockett's examples, as well as other examples randomly found in a dictionary, we would like to believe in a certain parallelism between double-stressed phrases and their counterparts of altered accentuation. Unfortunately, we would fail. Not only is always the first element stressed in phrases endowed with specific meaning, e.g., 'red brick (as referred to universities) vs. a 'red 'brick (i.e. a brick of red colour), or a 'redbreast (i.e. a robin); sometimes it is the other element that carries the stress, e.g., Long 'Island, South Da'kota, New 'York. The latter ones are geographical proper names, and their accentuation is practically opposite to that of common noun phrases, such as a 'long 'island, 'South O'hio, a 'new 'book. And what is more important, even some genuine compounds are double-stressed, e.g., 'brown 'betty (a kind of dessert), or 'best man to denote 'a man who helps the bridegroom'. Hence it follows that it is not primarily the stress contour that determines a complex as a compound; just on the contrary, two elements are linked with a certain stress contour because they are in fact sensed as compounds. For all that has been said on the issue, compounds are believed to be best defined on semantic grounds. The idea was already hinted in the above -quoted Zandvoort's view, and it is described explicitly by Jespersen (1966: 137): "As formal criteria thus fail us in English, we must fall back on semantics, and we may perhaps say that we have a compound if the meaning of the whole cannot be logically deduced from the meaning of the elements separately". In this respect, the stress phenomena do not signal the meaning directly, but they merely mark the contrast between compounds proper and other junctions meaningfully indivisible. Yet again, relying one-sidedly on either formal or semantic criteria will not provide us with a complex picture. Jespersen's requirement, too, seems to be rather rigorous, and it makes it impossible to regard as compounds such complexes as white-aproned, Anglo-Saxon, wood-fire, black-and-white, etc. Namely, no one would stop to think about the proper meanings of the respective elements and compose the total meaning out of them. In other words, the meanings of the complexes are predictable enough. However, it is true that the seman tic kernels, out of which the complex is made up, are inseparable. This is shown clearly by Trnka (1967: 110). In his opinion, semantic indivisibility is believed to be the only criterion reliable to determine and define a compound. (1)
Taking no significant account of phonetic (mainly stress) phenomena, I intend to proceed a little further in order to cover under one common head, if possible at all, original compounds, compounds proper, as well as certain loose collocations, or whatever labels the complexes have been affixed. My primary, fundamental assumption lies in taking a compound as a phenomenon of the process of determination within both the structural and the functional part of a sentence referred to as noun phrase (NP). What I have in mind is a semantic matter, which is reflected on the grammatical level as the relation of syntactic dependence between the head (i.e. the basic manifestation of NP) and the attribute. Moreover, admitting the existence of correspondence between the semantic and the grammatical levels, which reveals itself in a close or distant affinity between the grammatical elements and the respective semantic categories, I can easily render NP as a complex consisting of a string of slots into which appropriate attri butes come so as to perform various functions in connection to the head. Leaving the issue of determination-predication relation aside, I zero in on determination (modification) only, by which I understand the well-known Jespersen's junction, namely a connection of two units by means of which a new unit arises, more complex and more precise in meaning. (2) And here lies the crucial point for further considerations.
It is obvious that within one NP there can be two or more attributes connecting to the head. Due to the linear character of human speech, the attributes have to be ordered in a certain way. As I proved elsewhere, the degree of linkage of a respective pair in junction, on the one hand, and the position proper of the attributes, on the other hand, are interdependent phenomena. It holds that the closer is the relation, the nearer is the attribute posited to its head. Therefore I assume that the attributes of one and the same NP move, as it were, along a scale of onomasticity, on which the zero point represents a moment of "pre-speaking" and the first possible minimal value the loosest junction of the attribute and its head. The maximum value on the scale, on the other hand, indicates the closest junction possible, with a new, yet perhaps occasional, naming unit coming into existence. Within the pale of the two terminal values we are allowed to speak of something that may be referred to as onomasticity distributi on (OD). Briefly speaking, OD principle is believed to express a degree of naming value existing between the given attribute and its head. Now we may be sure that the first minimal degree of onomasticity is displayed over the junction in which the first element performs grammatical rather than semantic function (i.e. articles and kindred expressions, such as determiners). On the other hand, the maximum degree possible is to be observed in junctions which undoubtedly are genuine naming units, both in- and outside the given context. In that sense I can define a compound as a unit, namely a combination of (an) attribute(s) and the head which complies with the proposed OD principle in that the degree of onomasticity is felt as the highest.
The advantage of the approach described here above is believed to lie in the fact that English compounds are not treated as isolated phenomena of grammar and/or semantics but always as parts of concrete utterances. It should be noted that the degree of onomasticity in two (or more) semantic kernels within a NP is sometimes allowed to vary as the case may be. In other words, the very same semantic kernels being in the relation of determination may represent a compound in one NP whereas they are not a compound in another NP. In the latter case the junction is no longer a naming unit, which is practically reflected by the fact that another semantic kernel (i.e. an attribute) can intervene. In this sense we can render quite satisfactorily such NPs as a baby blackbird and a black baby-bird, or an evening paper and an evening radical paper, and the like. Context-sensitive compounds are also junctions young/old + one/fellow/ man/boy: while the degree of onomasticity displayed over these kernels is felt as the highes t in a sick young boy, a modern young lady, a handsome old fellow, the mere swapping of the attributes or an occasional insertion of another attribute is the result of a lowered degree of onomasticity. This will lead to more or less different meanings: thus a young modern English lady is not exactly the same as a modern English young lady, and an old handsome fellow says something on the fellow's advanced age, which does not necessarily have to be the case in a handsome old fellow. Likewise, fairly different meanings have the following NPs: burnt-out peasants' hovels vs. peasants' burnt-out hovels. Without gathering more illustrations at the moment, I point to the fact that any breach of the ordering, mainly of the ordering of the attributes closest to the head, is believed to be functional, and it can even result in rejecting the NP as meaningless, e.g., the white sweet pea, but not really * the sweet white pea.
How does the concept of compound comply with the theories of idiomaticity? It is true that compounds are mostly omitted in works written by scholars involved in idiomaticity. Some, like Sonomura (1996), do introduce the concept, though as one of the so-called "cross-cutting terms", besides binomials and phrasal and prepositional verbs. Examples that are usually presented do not cover the whole scope of complexes traditionally referred to as compounds, though. Such instances as blackbird, afternoon, statesman are missing, and so also are those that border on derivation, perhaps with the exception of N+N phrases, e.g., fellow- worker. It is no wonder because, on the one hand, scholars wish to be as delicate in classification as possible, but on the other hand, they must observe one of the fundamental postulates of any definition of idioms (and kindred expressions). As this refers to something like "multiword chunks" (or "concatenations"), consisting of minimum two words, the blackbird-type of compounds must be ignored, and the more so statesman, gospel, and only the fellow-worker or home nurse-type seems to be worth noticing.
Mine is the modest conviction that most, if not all, traditional compounds can be treated as idiomatic, surely in the broader sense of the idiomaticity concept. Moreover, and this is important, indeed, certain features typical of idiomatic expressions are characteristic for compounds, too.
At first sight, compounds look like free combinations, i.e. the standard type of meaning construction complying with the rules of syntax and being absolutely logical. Grammatical privileges are but minor, concerning practically plural formation. While in the overwhelming majority of cases the plural marker is attached to the whole complex, e.g., blackbirds, head- masters, home nurses, Frenchmen, it is idiomatic to say sons-in-law rather than son-in-laws (3) (though we have, quite expectedly, forget-me- nots). It is also worth of note that we say fellow-workers, assistant professors or alligator shoes, but women doctors rather than woman doctors. (Or is it because, and only because, woman doctors would rather mean 'gynaecologists', as I mentioned above?)
Although such "exceptions" do exist, they do not by themselves speak against the free-combination status of most compounds. The situation will change, however, once we begin to raise semantic issues. Of course, meanings of most compounds are fairly predictable, namely, their total meanings are derived easily from the meanings of their respective elements. Thus country folk, for instance, can hardly be understood in any other way than 'people living in the country, not in town', head-master is definitely 'a (male) teacher in charge of a school', and stone wall will not be made of plastic! Yet there is something about the semantics of compounds that speaks in favour of placing them beyond the sphere of casual free combinations.
First of all, meanings of some compounds are not predictable (in which Jespersen was right), and even if they might spring readily to our mind, they could sometimes be misinterpreted. At the beginning of the present article I introduced the issue by the example blackboard. So also blackbird: the meaning of this compound could by mistake be described as 'any bird of black feather', yet, those who are knowledgeable of the English vocabulary will understand it as a species of birds (whose females, by the way, are grey in colour!). The very expression was chosen by convention -- actually, a great many other species of fowls are black as well. For good measure, the equivalents to blackbird in other languages do not refer to those fowls' colour at all: let us compare Czech kos, German Amsel, French merle, Spanish mirlo, Swedish koltrast. Hence we can assume that the expression blackbird is a kind of idiomatic expression: (1) conventionality certainly applies to all idiomatic expressions, and since blackbird refers by convention to only one certain species and not to 'any black bird', it must, by that very token, be idiomatic; (2) it can even be compared to an idiom proper, this conviction being based on the definition of idiom, commonly accepted, which stresses the fact of impossibility of verbatim translations to other languages. And, indeed, Czech cerny ptak, German schwarzer Vogel, French oiseau noir, Spanish pajaro negro, or Swedish svart fagel gel have different meaning from the English blackbird.
A similar example is the White House, with one significant exception, namely, the reference to colour. And yet, its meaning is not predictable without the speaker's awareness of its working as a unique proper name, the name which was bestowed on the subject, and the institution, by convention. (4) Hence it also follows that the expression the White House appears to be indivisible, and the issue of compositionality is out of question as many other houses are also white and yet they are not the President's seats. Or if we insisted on this characteristic to be described, then the meaning of the White House would have to be best viewed as non-compositional, or perhaps only semi-compositional. Again, all the more reason for compounds to be treated as idiomatic expressions.
Subtle analyses and sophisticated consideration would probably allow us to recognise a kind of idiomatic character of the compound afternoon. The expression by itself is indivisible, and not fully predictable. Namely, the issue of compositionality cannot be answered satisfactorily, even if it may seem at first sight that the total meaning is composed, or derived from the meanings of "after" and "noon", thus referring to the latter half of a day. However, the length of that part of day cannot be predicted: not only do we have the expression evening in the English word-stock, but neither can we say very well when the afternoon period ends and when the evening one begins. It is just a matter of convention that specifies the spell of afternoon as a period between midday and sunset, or between midday and the end of day's work. (5)
I have so far been dealing with such types of complexes that are called compounds by tradition, to a certain extent ignored by scholars interested in idiomaticity, though. It remains to render those that comply better with the descriptive characteristics of an idiom(atic expression), e.g., lady-killer, alligator shoes, women doctors, and similar. Nevertheless, are these NPs examples of one and the same type (and the only one as mentioned by Sonomura), or are they representatives of several types? And do all of them have the right to be called compounds, in the light of idiomaticity theory?
The complexes illustrated here above seem to represent a kind of spectral differentiation, showing characteristics that are worth some comment. On the one end of the spectrum there seems to be alligator shoes, at first sight easily predictable, since its total meaning could be viewed as a composition of the meanings of "shoes" and "alligator", respectively. As for the first element, it is by experience that speakers will understand its semantics as "material", i.e. 'alligator's skin turned to leather'. Of the same type would then be, for example, stone wall, gold coin, and similar. Here, too, their meanings are compositional, and also fairly predictable. And yet we somehow feel a difference between alligator shoes and the latter two. Whereas "stone" and "gold" are referents to materials by themselves, and so the way of predictability is direct, "alligator" refers to material "metaphorically", as it were, which must be supported by convention. Therefore stone wall and gold coin are very close to free combinati ons, while alligator shoes tends to behave like an idiomatic expression. And not only because of that! A native speaker will surely understand a secondary, truly figurative meaning of alligator shoe, namely 'a shoe of a loose sole'. Besides others, this example speaks in favour of prevalence of semantic considerations over those within the surface-structure syntax.
The type lady-killer seems to stand at the opposite end of the supposed spectrum. The meaning of this complex is only partially compositional since the meaning of "killer" here is not regarded as literal. The intended direct predictability would then misguide us towards the interpretation 'a man who kills ladies'. This, of course, is not the case. By virtue of a figure, frequency of its use and fixation, the expression is understood conventionally as 'a man who charms and attracts all the women he meets' (6).
Finally, the type women doctors. No wonder that Sonomura presents this "compound" in plural -- its singular form is meaningfully ambiguous. If someone says, or, better, writes (7) Did you meet the new woman doctor?, we may feel uncertain whether the new doctor is a female or whether he means 'the new gynaecologist'. As I hinted here above, the ambiguity is avoided in plural: two plural markers refer to female doctors, while only one marker, attached to the whole of the expression, denotes specialists in gynaecology. It is rather difficult to assess the aspects of compositionally and hence of predictability. It seems, however, that the meaning of the expression is likely to be predicted easily, but what comes first to our mind is very probably the meaning 'female doctor'. The other meaning, i.e. 'gynaecologist', is only secondary, coming into existence through convention based on the familiarity of use. In either case, as I assume, woman doctor, being an indivisible expression, can be rightly called a compound .
In the light of idiomatic theory the status of compounds must be discussed and, hopefully, solved on semantic grounds. As elsewhere in language, such a procedure will be preferred that language phenomena are observed not as isolated facts but rather as interactive ones, allowing for a great deal of relativity of view. If we did not respect this approach, the concept of compound would, with all probability, get shrunk to a very narrowly defined category, something that Jespersen once proposed. (8) Such a rigorous view might lead us to a blind alley: a compound would then appear to be a mere synonymous term for "idiom proper", and thus many complexes traditionally referred to as compounds could not be taken into account. As I am trying to show, mine is the idea of transition, or "fluidity", which in terms of idiomaticity allows for the proposed relativity in such aspects and criteria as compositionality, predictability, variability, and even conventionality. In other words, I am convinced that all these aspects and criteria can be expressed in degrees, which vary from case to case. And here the proposed variability in degrees makes it possible for us to regard as compounds any complexes from the type blackbird, via the type afternoon, to the type country music. All these, as discussed above, are believed to be endowed with certain degrees of features characteristic or typical of idiomatic expressions. The degree of predictability in country music is certainly higher than in blackbird, but, on the other hand, lower than in country town. As a matter of fact, it is the degree of conventionality, namely familiarity of the specific meaning, that plays the decisive role here: no matter whether the given complex is compositional or non-compositional, once it comes into existence on conventional grounds, it becomes practically an indivisible expression. Therefore, what we understand by country house, for example, is not simply a building constructed outside a city to live in, as might be predictable from the meanings of "c ountry" + "house", but rather (and mainly for the British) 'a large house in the country, often of historical interest, and very likely open to public'. So also country music (equivalent to country western) will not be just any kind of music performed in the country or by country people but rather 'a specific genre of music, namely in the style of the southern and western United States'. And similarly, country club will say more than the information on the location of such a club, as considered compositionally; the expression is conventionalised to understand that one must be fairly rich to afford to become a member of a country club. All these illustrative cases, and many more, are characteristic of semantic indivisibility, which seems to be the most decisive criterion to claim for a complex to be called a compound.
I may not yet have stressed adequately enough that the status of compounds is generally dependent on, or sensitive to, context. As a matter of fact, one aspect of context was shown, at least illustratively, with respect to the OD principle: suffices to consider the change of meaning in a pretty young lady and a young pretty lady. From the other way around, we have to admit that the complex the white house will have a different status in the utterance May I take a picture of the white house?, as pronounced in front of the US President's residence, or somewhere in Paris or Prague in front of a white-plastered attractive historical building. Of course, idiomaticity in the sphere of phonetics requires that the two interpretations of the white house be pronounced in two different stress-intonation contours: the metonymic use of the white house in the illustrative interrogative sentence (spelt "the White House") will normally be characteristic of rising intonation in the stressed "white" (nucleus) and the unstresse d "house" (tail), whereas the ordinary free combination white house will show the expected rise in the stressed "house" only.
Similarly, if a wife expecting a baby says to her beloved husband that she must go and see her woman doctor, what she probably means, and what her husband will understand, is her 'gynaecologist', even if he may have misheard the stress lay-out expected for the "figurative" use (i.e. 'woman doctor). And it is also the context that will qualify red tape for either a genuine compound or a mere free combination in such utterances as Before we finally started repairing our country house, we had to cut through the red tape of all sorts of regulations and A red tape rather than a black one would be better to insulate the wire with for people to take care. Here, too, different accentuation accompanies the two meanings, giving preference to red 'tape for its idiomatic, figurative interpretation. However, what I particularly have in mind here is the fact that the semantic and, consequently, syntactic indivisibility of the junction red tape in the former case is apparent. We can say: to cut through the tiresome red tape , yet hardly * the red tiresome tape. On the other hand, another attributive element is allowed to come in between "red" and "tape" in the latter case, if, of course, this is in compliance with our OD principle; e.g., a red sticky tape. Then, in such a context, the complex sticky tape can be considered a new compound (synonymous to 'adhesive tape' or 'Scotch tape').
Indeed, we must always observe the context in order to be able to decide about the status of respective complexes. Whereas the true meaning of blue story, for instance, will probably be unpredictable in any reasonable context, its meaning being definitely non-literal, we shall need a fairly extended context to learn the proper interpretation of, say, blue hotel. There are no priori obstacles to prevent us from understanding a blue hotel as either one painted blue, literally, or one of nasty or possibly of mournful reputation (like in blue story, or blue mood). The latter meaning, unlike the former, is unpredictable, and it must be conventionalised. If this is the case, such a variation as a blue seaside hotel for the figurative meaning is excluded, for the same reason as * a blue long story is unacceptable in its only non-literal interpretation. Then blue hotel may become a good candidate of a compound.
Certain privileges in language creativity (as shown above by the example blue hotel) are only to document the fact that trying to squeeze language phenomena, compounds included, to neat sharp-cut boxes is sometimes impossible and therefore an improper procedure to follow. I hope to have managed to prove that the complexes traditionally labelled "compounds" can, indeed, be called so in the light of idiomatology, provided, of course, that the complexes in question are semantically indivisible within respective contexts. Namely, they must show at least the minimum amount of aspects, or characteristics, typical of idiomatic expressions. There will certainly be such complexes that have the full right to be referred to as compounds, and others which border on free combinations. The latter ones are compound candidates, awaiting, as it were, to be turned to genuine compounds, namely to such complexes that are semantically indivisible in any context imaginable once their new, specific meanings have been agreed on conv entionally by all native speakers.
(1.) Czech Anglicists will remember Mathesius, who also had a say in the issues (as early as 1910): his slozeniny (unlike souslovi 'workgroups' and sdruzeniny 'collocations'), i.e. 'compounds', were defined on absolutely grammatical (formal) grounds, though.
(2.) This is what also Kopecny (1962, [section] 51) understands under the term determination.
(3.) The latter is, phonetically, thc Saxon Genitive form, spelt correctly son-in-law's; it is interesting, that some people use it as nominative plural, though.
(4.) the White House is a typical example of metonymy.
(5.) Different periodisations of a day can be observed with speakers of other languages, but this has next to nothing to do with the issues of idiomaticity of the English compound afternoon.
(6.) Quoted from Longman Dictionary of Contemporary English (1978).
(7.) As we pointed out here above, the two meanings are differentiated on the phonetic level through stress contours.
(8.) See here above: Jespersen (1966: 31, 47, 85).
1933 Language. New York: Holt, Rinehart and Winston.
Gleason, H. A.
1965 Linguistics and English Grammar. New York: Holt, Rinehart and Winston.
Harris, Z. S.
1951 Methods in Structural Linguistics. Chicago: Chicago University Press.
1966 A Course in Modern Linguistics. New York: The Macmillan Co.
1966 Essentials of English Grammar. London: George Allen and Unwin Ltd.
1962 Zaklady ceske skladby Prague: Statni Pedagogicke Nakladatelstvi (SPN).
1910 "Poznamka a substantivnich slozeninach a sdruzeninach vsoucasne anglictine", Sbornik filosoficky 1.
Sears, D. A.
1972 "The system of compounding in Modern English", Lingua 91.
Sonomura, O. M.
1996 Idiomaticity in the Basic Writing of American English. New York: Peter Lang Publishing.
1967 Rozbor nynejsi spisovne anglictiny II. Prague: Statni Pedagogicke Nakladatelstvi.
Zandvoort, R. W.
1966 A Handbook of English Grammar. Bristol: Longmans.