2,657
Views
22
CrossRef citations to date
0
Altmetric
Original Articles

An Introduction to the Analysis of English Noun Compounds

Pages 356-373 | Published online: 04 Dec 2015

  • After this article was written appeared the work of Hans Marchand, The Categories and Types of Present-Day English Word-Formation (Wiesbaden, 1960). Professor Marchand mentions, casually, a number of Jespersen's notional types (p. 22), but is much more outspoken than the latter as to the impossibility of classification, going so far as to express indifference to the exact meaning of a compound.
  • It is curious that part of the evidence he adduces in support of his attitude consists of the fact that a given first-word (e.g. finger-), as it combines now with one, now with another second-word, may show a different logical relation with the latter: finger bowl, fingernail, finger post, fingerprint. As if any (uninfected) noun could bear within itself its own relation with something else!
  • Jespersen's description of his first category is awkward and inaccurate; it is surely not true that when the second word represents an agent-noun the first part of the compound may indicate the subject; when B is an agent-noun, A can be only the object: sun-worshipper.
  • (Since the type sun-worshipper may be seen as an extraction from the type sun-worship, I shall speak of the “verbal types” as if consisting only of sunshine, sun-worship.)
  • To his list of compounds of Purpose, Jespersen adds football, when this refers to the ball itself (=‘a ball for the foot’); when, however, it refers to the game, he believes it should be classified under Means. Here, surely, Jespersen is napping.
  • It is a frequent practise to name a game after the principal object used in playing it (e.g. quoits); this is what happened to the simple ball, as also to its compounds football, handball, softball. We have to do in such cases with a Bahuvrihi extension of the compound: football would come to mean ‘that which is characterized by a football: the game in which a football is used’. Obviously, in such an exocentric use, the second element still has the same relation as before with the first element.
  • Jespersen, however, appears to believe that we still have to do with an endocentric compound, in which the relation between A and B would have changed: the second element would now refer to the game, and the compound would mean: 'a game (of ball) played with the foot' (=Means). Did he also believe that softball, in its extension, must be interpreted: ‘a game (of ball) which is soft’?
  • If Jespersen had looked for a Time-to-which, as parallel to Place-to-which, he probably would not have found it. This may be partly explained by the fact that Place-to-which itself is not very frequent (and Time has the habit of imitating Place). One may note in this connexion, however, certain formations in which A designates a part of the body to which B “comes”: knee-pants, waistcoat; hip-boots, trunk-hose and, probably, breastwork (‘a defensive work of moderate height…’), and breast-rail (the upper part of a balcony…'). In this type, which seems to be improductive today, does A represent a place? If not, we have found a new sub-type: “A=Object to which B comes, reaches”. I prefer, however, the spatial interpretation, since the bodily part is taken as a point by which extension is measured.
  • Perhaps it was because Jespersen failed to note the specific relation of minute-hand that he was totally unable to account for weathercock, in which B serves the same function of ‘measuring, indicating’. Weathercock is one of a dozen compounds that Jespersen listed, at the end of his treatment, as representative of the many that “do not fit in anywhere.” A few of these formations are so obvious that it would be slightly embarrassing to “explain” them; for others, a look at the NED suffices (almshouse originally meant simply ‘a place where alms were kept’; cf. ice-house); others would require a serious historical investigation; finally, some of the compounds can be understood only against the background of more contemporary material than Jespersen evidently had at his disposal. Because of the unevenness of the problems involved, it seemed best not to take up in detail Jespersen's list of “unclassifiables.”
  • One might also ask: “where are shell-shóck and hay fever? fever blister and heat scars? inkspot and watermark?—that is, examples of A=Agency”? Jespersen's category 5, in which A=”tool or instrument” (of deliberate activity), would hardly cover these examples of “accidental result caused by A.”
  • Incidentally, it may be noted that while Jespersen used the formation gas-light to exemplify all types of determinative non-appositional compounds, this compound itself is nowhere classified. Would he have put gas-light under (5), next to foot-step? This would be no worse than marrying mountain range with newspaper. If gas-light means, as it usually does, ‘light from a gas-burner’ then its true congeners would be lamp-light and candle-light; and, very close, firelight, starlight, moonlight, sunlight; sunbeam, moonbeam. Where should we locate this “emanation” type, under Jespersen's Place-from-which? As a matter of fact, his land-breeze would fit in not too badly. There is a point at which Place-from-which and Thing-from-which seem to fuse, both of them drifting slowly in the direction of Agency.
  • One may also find cases of “B=Activity” for Jespersen's (4) and (6): fire-drill, sneak raid”, as for (5), he has allowed all his examples to show B as action-word.
  • It must be admitted that Jespersen's desire to keep the sunshine, sun-worship types separate from all the rest of the compounds with B=action-word, is understandable. They do seem to represent a special case, though it is not too easy to state precisely why, and often difficult to distinguish them. If Jespersen had explained this, if he had taken some pains to justify their being treated as a special type of compound within (or not quite within) the gas-light type, then he would have been in his rights in presenting them “in their own terms.”
  • Jespersen was, for example, much more interested in classifying the suffixes of English phonetically: to this he devoted 150 pages.
  • Mätzner clearly points out the relation underlying the type castor oil (A=”… der Gegenstand, aus und von welchem etwas gewonnen wird oder entsteht…:” I, 527); while we find no specific mention of the relation “B is a part of A” (broomstick), this type is at least represented in his category of Genitival relationship (e.g. shipboard).
  • That Jespersen overlooked the exceedingly important shipboard may perhaps be explained as follows: in the miscellany represented by Mätzner's Genitival group, there appeared also examples of the verbal type sunshine; Jespersen extracted this one particular type, expanding it slightly, and placed it in direct connexion with its opposite sun-worship (illustrated by Mätzner under “Accusatival relationship”), in order to construct his category 1. Not only did Jespersen overlook the logical flaw involved in setting up this category without warning (note 7): he apparently forgot to go back to Mätzner's Genitival group and rescue shipboard. That he never sensed its absence is strange.
  • To illustrate Place-to-which, Mätzner offered side-glance, church-goer; with the more worldly Jespersen, we find side-glance, play-goer. (Incidentally, do not both church-goer and play-goer belong to the verbal type of category 1? Cf. also streetwalker.)
  • It should be insisted upon that the main categories of Mätzner (which he constructed to include also appositional and copulative compounds) were logically flawless. He has two main headings: “The two elements belong to the same ‘case’” (apposition; copulative cps.) and “The two elements belong to different cases.” Under the second, he distinguishes between those compounds in which A may be considered a direct case of B (Genitival and Accusatival relationships) and those in which A's relation to B can be explained only by recourse to a preposition (thus A would represent a “prepositional case,” with the preposition missing); it is with the prepositional types, obviously, that he places the six other categories we know from Jespersen. Though his ruling concept was the unwieldy one of “case,” still Mätzner had a ruling concept, and Jespersen had none.
  • It must be admitted in Jespersen's favor that one of his suppressions was well-inspired. Mätzner had included, along with Characterizing Feature, Material, Means and Purpose, a fifth type: “A=that with which a Person B concerns himself or works:” ale-wife, blacksmith etc. This category not only offers a blend of reference and relation but is superfluous, being a subdivision of Purpose. Jespersen deleted the category as such while borrowing certain examples of the blacksmith type for a comparison with the shoemaker type, which he offers in a comment on his category 1.
  • As for Koziol (Handbuch der Englischen Wortbildungslehre, Heidelberg, 1937), he offers no improvements on Mätzner; his only change (apart from simplification) was to omit entirely Mätzner's few subdivisions of Place and Time, presenting these two unsifted references as “Spatial and Temporal Relationships.” Was this his way of bridging the gap between the two levels of criteria?
  • What a singular lack of interest in the semantic problems of English compounding has marked the century after the appearance of Mätzner's classification! No one has attempted to carry his results further, nor has he been challenged (except by Nils Bergsten who, in his thesis, A Study on Compound Substantives in English, Uppsala 1911, attempted to replace Mätzner's categories by others based on the theories of Noreen). An even greater indifference seems to have characterized German scholarship; at least, in 1939, John T. Carr (Nominal Compounds in Germanic, Oxford) was able to say that, with the exception of one brief article, no contribution to the semantic analysis of Germanic compounds had been offered since Grimm! Perhaps the century-long stagnation is to be partly explained by the fact that for some decades the energy that might have gone into a re-investigation of such problems was devoted to endless debates on defining the compoundness of a compound.
  • If the lesser quantity of A contained in AB is actually the criterium for distinguishing Characterizing Feature from Material, it is odd that Jespersen has included in (6), by the side of sandpaper, the compound mountain-range: what does a mountain range contain besides mountains?
  • Under Characterizing Feature Jespersen offers no examples of the 3–member type of compounds that has become so frequent in the last few decades (at least in America): kick-pleat skirt, shawl-collar dress; gate-leg table, roll-top desk; 2–button glove, 4–motor bomber; full-size sheets, full-length portraits; high protein foods, high-proof whiskey; 2–foot pole etc. (See my article in Modern Language Notes, December 1944). Jespersen does offer in (2): all-night club; but one may wonder whether all-night is not a (compound) adverb rather than a (compound) noun.
  • The reader will surely understand that it was only for purposes of quick demonstration that I purported to “derive” my set of categories from those of Mätzner and Jespersen—as in a kind of parlor game. Actually I arrived at it, very slowly, basing myself on the analysis of a rather large number of examples; it is because I found my classification capable of accomodating all my material that I make bold to offer it here.
  • Note that OE abutan <on butan meant ‘on the outside of’, i.e. ‘containing’.
  • As concerns the particular formation newspaper, which to Jespersen was a paper ‘containing’ news, I prefer to see this, like trade journal, under Purpose: ‘B is for, B is devoted to, A’. I believe, that is, that in both cases we have to do, primarily, with a journalistic institution. (Incidentally, I have not pointed out all the cases of possible mis-classification by Jespersen.)
  • If the category of “concrete condition” is valid, it very probably should include much more than has been indicated above (I shall suggest later that ice may be a condition).
  • As for Concrete Object (and, occasionally, Substance), it may often be very helpful to make special subdivisions for (a) parts of the body and (b) plants.
  • The terms Object and Substance, themselves, correspond more or less to Thing-Word or Countable, and Mass-Word. Plurals, as may be imagined, are often troublesome; sometimes they seem to go with Substance (cf. asfoetida bag, lea-bag, sand-bag, and bean-bag=‘a bag full of beans’; incidentally, Jespersen does not recognize, as Mätzner did, the sub-relation “B is full of A”); sometimes it is as if plurals represent a third possibility.
  • (I realize that to suggest a number of subdivisions for 3 and none whatever for the infinitely complicated items 5, 6, 7, is highly disproportionate.)
  • I am classifying the flu of flu virus under 7, instead of under 6, where some might prefer to place it (since a disease may be thought of as a process).
  • One will note (probably with some mental discomfort) that within the different groups just listed, the logical relation shifts at random. Even if I had desired, I probably could not have kept it constant, since not all of the referential types can be found in all of the main categories. A more complete presentation would have included indications of the distribution of the 49 referential combinations. This I have not yet worked out to my satisfaction. But it would seem that while a few of the types belong to only one of the main categories, and a few others (including, obviously, A3B3) go through all four, the average is slightly over 2 relational categories for each compound reference. This would give us, then, a good 100 subdivisions.
  • A particularly interesting example of the influence of reference upon relation has to do with the two types pointed out by Mätzner-Jespersen: Characterizing Feature and Material. As we have seen, these were presented as main categories, apparently distinct the one from the other, and of equal weight. Actually, of the 49 referential subdivisions that can theoretically appear in Ⓐ (and a couple of dozen may appear), only one can accomodate both relations: only with A3B3 (sand paper, wool dress) can one speak of Characterizing Feature and Material; with all the rest, Material is excluded. A3B6 gives something like dress rehearsal; A3B2 wart-hog; A3B1 white-collar worker; A6B6–7 whooping-cough; A6B1 yes man; A7B6 labor dispute; A7B3 first-class material etc. Moreover, not all of A3B3 can give Material, only A3bB3a, b (wool dress, wool material); with A3cB3 we get the type loaf sugar, ball fringe, print butter, plug tobacco; with A3aB3a: motor boat, machine gun, bagpipe, brake drum, slide rule. Thus, only a certain part of one referential combination gives the “main logical category” of Material. Moreover, when this sub-relation does appear, it may fuse indistinguishably with Characterizing Feature; compare, for example, with foods: cinnamon bun, lemon pie, chocolate pie, apple pie, deep-dish apple-pie, batter bread (where A= 100% of AB); at what point has our cook passed from one logical category to another?
  • It is only within referential limitations that meaningful statements can be made about the sub-relations and their inter-connexions. A3B3 should be studied separately; the “Wörter und Sachen” problems that may arise here (and I have pointed out only one of a number of similar problems) will have practically no parallels in the other 48 compound references (occasionally, A2B3 will remind us of A3B3).
  • I speak of the difficulty of distinguishing between references and between sub-relations; I have said nothing of the difficulty of distinguishing between the 4 main relational categories. Can one always recognize Ⓐ, Ⓑ, A→B and A←B? I should say that if we understand the meaning of our compounds and if we have a decent corpus of examples, we will, mainly, not have to hesitate. And in the few cases where choice appears difficult, we may find that the difficult types come in great blocks. For example, consider mountain-range, and the many other compounds in which B refers to shape, mass, extension: land mass, sand dune; sandpile (ragpile etc.), haystack; shell mound, dung hill; melon balls, snowball, snowflake (soap-flakes), snowdrop, tear drop, chocolate drop, butter pat, sugarloaf and many, many others (a type never before isolated, to my knowledge). Shall we assign these to Ⓐ, as representing subtle variants of Material? Compare smoke ring and gold ring; snowball and rubber ball; sand bar and iron bars. Or perhaps they belong under Ⓑ; perhaps in snowball the real entity is snow, and ball refers only to an aspect of this entity, to its shape; must not an entity contain its own shape and not the reverse? And if sugarloaf, for example, is placed under Ⓑ, it will offer a perfect contrast to loaf sugar (print butter, etc.), which can only be interpreted as Ⓐ.
  • But is it really true that in snowball the second element represents only an aspect of the first? Is it not more natural to think of ‘a “ball” containing snow’ than of the reverse? Perhaps, after all, snowball and sugarloaf belong under Ⓐ. This would mean, of course, that Ⓐ must then accommodate both sugarloaf and loaf sugar—an apparent absurdity. How, how, shall we classify sugarloaf and snowball, these reminders of the philosophical problem of Entity and Aspect? Whatever the answer, I know at least that we can find hundreds of examples of the easily recognizable, snowball type: it is a “great” type. Thus we should not worry overmuch; the great types abide our classification; our first task is to isolate them.
  • Another problem which may have occurred to the reader: will not the procedure of using referential limitations as a guide to sub-relations, lead, if rigidly carried out, to over-classification? It will surely mean that we must separate water-power from horsepower, because a horse is a horse; we must separate also hay fever from trench fever. Is this not absurd? I think we will know only at the end whether or not it is absurd; if it is, the flaw can be rather easily remedied. But a method of classification which is only semi-systematic can never be remedied. And we should not appreciate too highly our ability to distinguish such obvious similarities as those connecting horse power and water power, trench/ever and hay fever; what each pair has in common, any idiot can see. Perhaps the language has other analogies, less obvious, more consequential, to show us.

Reprints and Corporate Permissions

Please note: Selecting permissions does not provide access to the full text of the article, please see our help page How do I view content?

To request a reprint or corporate permissions for this article, please click on the relevant link below:

Academic Permissions

Please note: Selecting permissions does not provide access to the full text of the article, please see our help page How do I view content?

Obtain permissions instantly via Rightslink by clicking on the button below:

If you are unable to obtain permissions via Rightslink, please complete and submit this Permissions form. For more information, please visit our Permissions help page.