Skip to navigation – Site map

Semantic Complexity in English [NN]n Compounds

Pierre J.L. Arnaud
p. 7-21

Abstract

Cet article passe en revue les complexités de la sémantique des composés [NN]N anglais, depuis les diverses combinaisons dénotationnelles entre N1 et Ν2 dans les unités non-endocentriques, jusqu'aux cas très compliqués où plusieurs tropes sur les composants et/ou sur le composé se superposent. Il examine ensuite quelques directions de recherche sur l'interprétation des composés en regard de cette complexité.

Top of page

Index terms

Top of page

Full text

1. INTRODUCTION, FIRST COMPLEXITIES

1English nominal compounding involves considerable compression of information resulting in a single [N1N2]N structure corresponding to a wide range of possible types of denotation and N1 – N2 semantic relations. Compounds can first be divided according to whether one of their components is a semantic head, i.e. is the name of the concept that dominates the concept signified by [N1N2]N, with N1 acting as a modifier restricting the set of denotata of N2 (the IS A condition of Allen 1978: 105). Since semantically-headed English compounds are right-headed, the test is as follows:

  • 1 A variant of this test,
    (a) [N
    1N2]N is a kind of N2
    is actually a test of hyponymy and may not return (...)

2(A) (a) [N1N2]N is (a) N21

The result is positive with compounds like

(1)

bathtub, car bomb, pant suit, concertina wire

but the resulting sentences are infelicitous to various degrees with units like

  • 2 Some of the examples are restricted to the British or North-American dialect zones; tractor-trailer (...)

(2)

judge advocate, tractor trailer,2 airhead, top dog

(3)

?a judge advocate is an advocate
*a tractor trailer is a trailer
*an airhead is a head
*a top dog is a dog

  • 3 See Noordegraaf (1989) for a history of endocentric before Bloomfield.

3Units with negative results on Test A and which consequently do not have N2 as semantic head have usually been labelled exocentric, a term popularized3 by Bloomfield (1933: 235-36) which causes some confusion when applied indistinctly to different semantic types of compounds (Bisetto and Scalise 2005). Indeed, negative results on Test A are not sufficient to define a class of compounds and the following test (B) produces different results on the same units:

4(B) (a) [N1N2]N is (a) N1 and (a) N2

(4)

a judge advocate is a judge and an advocate
a tractor trailer is a tractor and a trailer
*an airhead is air and a head
*a top dog is a top and a dog

  • 4 Or, if a clear hyperonymic lexeme does not exist (which is the case for singer and songwriter), the (...)
  • 5 Bauer (1998, 2006) discusses the limitations of using stress as a criterion of compounding.

5Examples like airhead and top dog will be discussed later. Judge advocate and tractor trailer result from the non-hierarchical juxtaposition of two nouns, and the term bi-centric would therefore constitute a better description than exocentric. An important characteristic is that Ν1 and N2 are co-hyponyms.4 Bi-centric units take the plural –s at their right end as shown by the majority of occurrences of judge advocates and tractor trailers present on the web, but this is more a proof of lexicalization than of compounding. They usually have the /21/ stress pattern, contrary to the general tendency of compounds,5 and there exist non-hierarchical, polycentric units like crawler-loader-dozer or coach-café-saloon (Kuiper 1999, Renner 2006), in contradiction to the normally binary structure of compounds, so their status is somewhat in doubt.

  • 6 Dvandva, as well as bahuvrihi (to be introduced further on), are terms borrowed from Sanskrit lingu (...)

6The first aspect of compounding complexity I shall be reviewing is found in the semantics of bicentrics, in particular that of the and in their paraphrase, and denotational, encyclopaedic considerations must be brought into play. Judge advocate denotes a military lawyer who can serve either as a judge or a defense lawyer according to his posting, so he can be a judge or an advocate. The denotatum of singer-songwriter is simultaneously both a singer and a songwriter. Tractortrailer is a different case as this word denotes a kind of truck consisting of the coupling of a tractor and a trailer, a tractor plus a trailer, which makes it an additive compound or dvandva.6 If the two components of a bicentric compound are on an equal semantic footing, is their form, i.e. the order of their components (judge advocate, not *advocate judge), due to chance or to some discernible factors? Recent research (Renner 2006) indicates that what is at work is a combination of the denotational saliency and phonological weight of the components. Some units, like child soldier, behave like combinatory bicentrics on tests A and B, but show some differences with them: child soldier is apparently stressed /12/ in isolation for some informants, and its plural form on the web hesitates between children soldiers and child soldiers. Semantically, child and soldier are not co-hyponyms, and we have a choice between relaxing the definition of bicentric compounds or creating another category, which was the solution retained by Bauer and Huddleston (2002: 1648) with their ‘ascriptive’ category, while Benczes (2006: 107) recommends grouping them with analogical compounds (see below). Taken together, then, the semantic complexity of bicentrics is in inverse proportion to their limited numbers.

  • 7 The American Heritage College Dictionary, 3rd ed., has bluefin tuna; COD11, however, has an entry b (...)
  • 8 Airhead obviously never denoted a head in the first place and it is actually a convenient fiction t (...)

7Following Bloomfield (1933) a unit like airhead is said to be exocentric because its head is implicitly external: in the same way as we have bluefin tuna,7 where the head is tuna, we would have razorback (pig). Another way of looking at this kind of compound is to consider that it involves a trope (a metonymy in the present example), in which case there is no need to posit the existence of an external head and the term exocentric is not appropriate. In addition, as we have seen, the term unfortunately conflates bicentrics with trope (metonymic or metaphoric) compounds, and these are two very different categories. It would not be a good idea to get rid of it, however, given its wide acceptance, but it can be improved on if we consider that trope exocentrics, if taken literally, do consist of a modifier and head (Kuiper 1999, Bauer and Renouf 2001, Søgaard 2004: Note 1): an airhead is at first a head which has air in it and a top dog is a dog which is at the top, and the trope causes a secondary shift in the denotation.8 For this reason, the term secondary exocentric seems adequate. We shall return to the subject of tropes later, and the next section examines compounds where both components are literal.

2. MODIFICATION RELATIONS

8English endocentric compounds consist of a modifier and a head, but behind this apparent unity lurks considerable semantic complexity. A test, however, separates two major classes fairly easily:

(C) (a) N1N2 is (a) N2 that is like (a) N1

Units like

(5)

catfish, pin stripe, stiletto heel, queen bee

pass the test, while others like

(6)

pipe tobacco, motor boat, (Indian) rope trick, snow flake, periscope depth

9fail. The like in the test shows that the modification in catfish, etc. is analogical, and it is worth examining this analogy in detail. A catfish is a fish that is like a cat in that is has whiskers, not in that it is a mammal, is furry, has claws, etc. A pin stripe is a stripe that is like a pin in that it is very thin, not in that it is sharp, made of metal, has a head, etc. The modification rests on a subset of the features of N1 that is sufficient to create a subclass of the denotata of N2. Features common to N1 and N2, like [animate] in the case of catfish probably facilitate the apperception of the analogy and make it more salient, but do not contribute to the modification itself, and differential features are suppressed. This is exactly what happens in metaphor, with a major difference, though: in metaphor, a trope, the source, is substituted for the target, while both are present in a [NN]N compound, so in praesentia metaphor is a good descriptive term for this category of relation (Arnaud 2003: 87sq.).

  • 9 Readers may accuse me of having selected a first set of endocentrics that test positive on C, while (...)

10Modification in the other examples cannot be tested so easily.9 For instance, definitory sentences could be:

(7)

pipe tobacco is tobacco that is for the pipe
a motor boat is a boat that has a motor/that includes a motor as one of its parts/that

is powered by a motor
a rope trick is a trick performed with a rope/a trick about a rope
a snowflake is a flake consisting of snow/constituting snow
periscope depth is a depth at which the periscope can be operated

11Other, more formalized ways of paraphrasing such compounds exist (see for instance Brekle 1970; Hansen et al. 1982: 45sq.), but the above sentences are close to those resulting from the application of test C to analogical compounds. The well-known problems of compound semantics are perfectly apparent on this limited sample: the number of different relations can be expected to be large, but depends to some extent on the granularity of the classification, and some units are ambiguous, or rather superimpose different relations. This superimposition of semantic relations, which has been termed promiscuity by Jackendoff (forthcoming), is an important characteristic of endocentric compounds and a major source of complexity.

  • 10 In the form of 13 underlying generalized verbs plus 6 relations that cannot be generated thus.

12Promiscuity, however, is only the co-presence of two or more relations that can be described separately, and the main problem, together with that of granularity, is consistency in labelling when setting up a list of relations from a collection of compounds. There has been a long history of attempts at classifying English compounding relations at least since Maetzner (1860). While some authors like Jespersen (1942: 142sq.) and Selkirk (1982) have judged the inventory of relations to be potentially endless, others have listed a restricted number of abstract relations: 4 for Hatcher (1960), 6 for Warren (1978: 47-48), 9 for Levi (1978), 11 for Adams (1973: 57sq.), 19 for Lees (1970).10 Jackendoff rightly mentions “the literature’s limited success at enumerating them”.

  • 11 I find it preferable to restrict the term synthetic to [NVer]N and [NVing]N units like circuit brea (...)

13The term relational compound, although far from perfect, has gained some currency as the label for this class. What these compounds have in common is not really semantic, but constructional. In fact, all may be conceived of as incorporating an underlying predication, and a component can correspond to an argument, a complement, or the verb itself if it is a deverbal noun as in water catchment or a verb base as in tooth decay.11 Compounding compacts the predication, leaving only two of its elements. In the same way as analogical compounds have an affinity with metaphor, non-analogical endocentrics can be considered as in praesentia metonymies, as, on the one hand metonymy also involves short cuts across a possible underlying predication and on the other hand its occurrences belong to many different categories whose classification is also a daunting task.

14Authors who have attempted to list categories of N1 to N2 modification types in endocentric compounds do not generally seem to have realized that the analogical type of modification is radically different from the others. Morphological support for this idea, however, can be found in French endocentric compounds among which analogical units are exclusively [N1N2]N (French compounds are left-headed), like

(8)

poisson-chat, homme-grenouille, talon aiguille, écrou papillon, chien-loup

contrary to the vast majority of non-analogical units which are prepositional, generally with the vague prepositions de or à:

  • 12 There is also a growing minority of non-analogical [N1N2]N units like cas sujet, code barre, capita (...)

(9)

bateau à moteur, flocon de neige, pomme de terre, arbre à pain, billet de banque.12

  • 13 For a similar analysis on guard dog, see Jackendoff (forthcoming).

15Some analogical units like spy plane, however, are less obviously analogical than others. In this particular example, this may be due to the fact that the first component is as much verbal as nominal in meaning, so we have two possible paraphrases: a plane that is like a spy/a plane that is for spying, a case of promiscuity resulting in an intersection between the analogical and non-analogical categories.13 Analogy may also be less salient when a large number of compounds include the same N1, which results in a partial semantic adjectivization of N1 which backgrounds the analogy, as in examples like mammoth N, blanket Ν or lightning N.

16After examining the metaphor-like or metonymy-like nature of modification links in endocentric compounds, we now turn to metaphor and metonymy operating on the separate components and then on the whole compound.

3. METAPHOR, METONYMY AND COMPOUNDING14

  • 14 Some of the material in this section was presented at the Journée d’études sur les composés, Univer (...)

17Any lexeme may be subject to a metaphor or metonymy, and since compounds result from the assembly of two lexemes, some of them are bound to include units that have been the object of a trope. To decide whether this is the case it is first necessary to make sure that the component concerned does result from a trope, and decision isn’t always easy: in watermelon, for example, does N2 denote a melon, or a melon-like fruit? Informants are generally hesitant. Careful examination of dictionary definitions of melon and watermelon shows that the watermelon is not a melon but a melon-like fruit of a different genus with a high water content, so we do have a metaphor on N2. Furthermore, if the trope exists outside the compound, this means that it probably appeared earlier than it and this is of no great interest for compounding theory. A criterion of independent existence is the inclusion of this sense as a sub-entry in dictionaries within the entry for N2 and presence in discourse outside the compound: this is the case in armpit, for instance, where N2 constitutes an independent anatomical metaphor, as appears in pit of the stomach. Examples with compound-independent tropes are:

(10)

mouse click
bird sanctuary
horse regiment
town square

(metaphor on N1)
(metaphor on N
2)
(metonymy on N
1)
(metonymy on N
2)

18Entrenched (catachretic) tropic senses on N2 produce acceptable utterances on test A:

(11)

A radar screen is a screen

19More interesting in a discussion of compounding are cases where the trope appeared alongside the compound’s formation. For instance, farm in salmon farm or wind farm may have a sub-entry at farm in COD11, but informants generally do not like the sentence

(12)

?A wind farm is a farm

20A similar example is fish as in jellyfish, shellfish, starfish, which has a subentry at fish in the same dictionary but cannot be used in isolation to refer to the marine creatures bearing those names.

21The following are examples of compounding-induced tropes:

(13)

flame tree, glass snake
ribcage, body clock
pulp magazine, skin flick
lipstick, shoehorn

metaphor on N1
metaphor on N2
metonymy on N1
metonymy on N2

Test A gives infelicitous results when the trope is on N2:

(14)

?a body clock is a clock?
?lipstick is (a) stick

22In the above cases, one of the components was subject to a trope while the other remained literal, so the trope may be said to operate inside the compound. In what follows, we return to secondary exocentrics. As we saw, these compounds involve modification of N2 by N1, but N2 is not the semantic head. Two cases need to be distinguished depending on whether the literal sense survives side by side with the tropic one or has been displaced by the trope (respectively non-exclusive and exclusive tropes). Note that in the case of non-exclusive tropes, a [N1N2]N word-form is part of a homonymous pair, one of the units being endocentric and the other one a secondary exocentric. Here again, some precautions are in order, in particular the close inspection of dictionary definitions and corpus investigations. Examples are

(15)

Keystone
price tag
pigtail
heart-throb

non-exclusive metaphor
non-exclusive metonymy
exclusive metaphor
exclusive metonymy

23It is important to note that, unlike the case of a trope on N2, it is the whole compound that is here the object of the trope: Pigtail denotes a hairdo which is like a pig’s tail, but it is not a real tail and is related to humans, not pigs. A heart-throb is a man and has nothing to do directly with the literal heart. Consequently, as mentioned earlier, test A returns negative results:

(16)

*a pigtail is a tail
*a heart-throb is a throb

24Among metonymic secondary exocentrics, a dominant category is that of synecdochic compounds like greyback or hornbill, to which the term bahuvrihi should be reserved.

4. MORE COMPLEXITY

25Perhaps the least degree of complexity is that of bicentric compounds, particularly dvandvas in which the juxtaposition of N1 and N2 is iconic of the adjacency of their denotata in the denotatum of [N1N2]N. Within endocentrics, we have seen that non-analogical modification is much more complex than the analogical type, especially when different relations are present in one compound, and tropes can add a further dose of complexity. Another source of complexity is the intersection of categories, with examples like the already quoted spy plane (endocentric, analogical and non-analogical) and keystone (endocentric or secondary exocentric). There are other sources of complexity.

4.1. Different tropes on the components

26The two components may have been separately the object of tropes, either independently, which, as said earlier, does not directly concern compounding, with examples like:

(17)

shadow cabinet

metaphor-metonymy

or in the compounding process:

(18)

iron lung

metonymy-metaphor

4.2. Internal modification + overall trope: combinations in secondary exocentrics

27As discussed above, the modification in endocentric compounds rests on in praesentia metaphors or metonymies. In a secondary exocentric unit, four combinations are therefore theoretically possible between the internal modification and the overall trope. Fig leaf “something used to hide an embarrassing fact” has non-analogical, merological modification (a leaf is a part of a fig(-tree)), exactly like cloverleaf “motorway intersection”. Both are secondarily metaphorized, the difference between them being that the metaphor in fig leaf rests on functional features (a representation of a literal fig leaf was used to hide the indecent parts of statues), while cloverleaf rests on perceptual analogy. Rust bucket “badly maintained ship” has a ‘substance’ non-analogical modification combined with an overall perceptual metaphor. Scumbag “unpleasant individual” combines the ‘contents-container’ modification with an overall metaphor. Interestingly, there does not seem to be in my collection of some 3,000 [Ν1Ν2]Ν compounds any unit with analogical modification and an overall metaphor, and this deserves further investigation.

28On the contrary, metonymic compounds apparently include the two types of internal modification. In paperback, we have the non-analogical relation of “substance”, and heartthrob manifests the seat/process relation, which is also non-analogical. The analogical type of modification, on the other hand, is present in many bahuvrihis mainly denoting animals or plants, like spoonbill, shovelnose, ironbark. Paraphrases bring the internal modification relation to the surface (the square brackets correspond to the overall trope and the round ones to the modification):

(19)

[a cheesewood is a tree that has (wood] that is like cheese)
[a fantail is a bird that has (a tail] that is like afan)

4.3. Successive tropes

29The recent renewed interest in metaphor and metonymy, particularly in Cognitive linguistics, has led to an awareness of the interaction of the two tropes, which can combine their effects on one unit (Barcelona 2000, 2003; Goossens 2003). Geeraerts (2003) demonstrated this on idiomatic phraseologies, and Warren (1992) and Benczes (2005, 2006) examined examples of compounds in this light. It is therefore another type of complexity that is to be found here.

30The noun Leatherneck “US Marine” comes from the leather collar, or more precisely the leather lining of the collar of a uniform worn by US Marines in the 19th century. N1 is literal and the modification relation is non-analogical (‘substance’), and N2 is independently metonymized, since the target is a part of a garment and the source the part of the body covered by that part, as in leg (of trousers). Finally, the name of a garment part thus denoted denotes in turn a soldier.

  • 15 In an adequate context bottom line could have literal reference, but this sequence is probably not (...)
  • 16 The difference with Leatherstocking, the proper name of the eponymous character of a series of nove (...)

31While in the case of leatherneck an overall metonymy succeeds a metonymy on one of the components, other units can be submitted to a succession of overall tropes. Bottom line first denotes by a ‘container/contents’ metonymy the contents of the bottom line, i.e. the balance of the budget.15 A metaphor then carries this forward to the non-financial sense “what counts (at the end of a discussion, etc.)”. The opposite order is found in leatherjacket, the homonymous names of an insect larva or a fish, whose tough tegument (which is neither a literal jacket nor made of leather) is metaphorically denoted by the compound which is then metonymized into a bahuvrihi.16

4.4. When a non-analogical modification is brought about by a metaphorized component

  • 17 We could probably add rattlesnake to this list, but my informants were unable to choose between a r (...)

32Fish names constitute precious material for the lexicologist and the following set manifests a different kind of complexity: swordfish, sawfish, sailfish, spearfish. We can add to these a bird’s name, lyrebird, which will show the specificity of this category by comparison with lyretail, another fish name.17

33Lyretail is a bahuvrihi, while lyrebird tests positive on A, the test of endocentricity. Unlike a unit like catfish, however, according to some informants who know its denotatum, lyrebird produces a slightly infelicitous utterance on C, the test of analogical modification:

(20)

a catfish is a fish that is like a cat
?a lyrebird is a bird that is like a lyre

34Lyrebird is better defined in

(21)

a lyrebird is a bird (that has a tail that is like) a lyre

while lyretail corresponds to

(22)

a lyretail is (a fish that has) a tail (that is like) a lyre

35In both cases, compounding removes part of the information (in brackets), but not the same part. Similarly, sawfish can be paraphrased by

(23)

a sawfish is a fish that has a rostrum that is like a saw

and sailfish by

(24)

sailfish is a fish that has a dorsal fin that is like a sail

36Note that in the case of sailfish the utterance resulting from Test C is somewhat infelicitous:

(25)

?a sail fish is a fish that is like a sail

37The compounds in this category, which might superficially be confused with analogical endocentric compounds, are actually non-analogical endocentrics (relationals) with a metaphorized N1. We are very close to them with eggplant, whose denotata originally had fruit resembling eggs in shape and colour, the difference being that the modification relation in this case is ‘producer/product’ while that in the preceding examples is merological.

4.5. The relation of typicality

  • 18 When N1 is [human], the tendency is for semantically similar compounds to be genitival as in busman (...)

38Analogy is subjectively present in units like bear hug, beeline, catcall, dogtrot, donkey work, goose step, horse laugh, wasp waist, monkey chants, kimono sleeve, jockey cap (“of a kind worn by jockeys” COD11), sailor suit.18 Such units produce acceptable utterances on the test of endocentricity (A):

(26)

a bear hug is a hug

and the modification is not directly analogical (Test B):

(27)

*a bear hug is a hug that is like a bear

39The following definition, however, is revealing:

(28)

a bear hug is a hug that is like the hug of a bear

40But after all, one could ascribe this behaviour to the fact that any conceptual category rests on analogy between its members. Were this true, however, we could expect all literal endocentric compounds to lend themselves to the same paraphrase pattern, which obviously is not the case:

(29)

*bubblegum is gum that is like the gum of a bubble
*a cat burglar is a burglar that is like the burglar of a cat

  • 19 Have” is a label for a non-merological abstract relation which could equally well be labelled “Of” (...)

41It would seem, therefore, that we have here a specific category of compounds in which the modification is complex and combines in a non-promiscuous manner analogy and a metonymic relation, “have”.19 For this I suggest the term typicality relation. This relation was identified by Adams (1973: 57sq.), who classified it as a sub-category (VII C) of her category VII, Resemblance, but the results on the test of analogical modification (B) shows that this classification was inadequate. As the existence of the typicality relation blurs the distinction between analogical and non-analogical endocentric compounds, it is another source of complexity.

42Adding yet more complexity, a unit with the typicality relation can be metonymized into a bahuvrihi. This is the case for bullhead, another fish name, whose definitory sentence is

(30)

a bullhead is a fish that has a head that is like the head of a bull

43and which incorporates three basic relations: the merological and the analogical ones, then again the merological in the metonymy. This is an extreme case, it seems.

44Finally, isolated instances of complexity arise in various circumstances, for example where there exists a N/Adj pair, as in the case of platinum blonde which can be paraphrased as:

(31)

a platinum blonde is a woman who has hair that is blonde like platinum

45The fact that the noun blonde metonymically compresses the information “woman with blonde hair” causes the analogical modification to apply directly to it, with a resulting hypallage.

5. DISCUSSION AND CONCLUSION

  • 20 OED and World Wide Words website.

46How can compounding theory deal with all that complexity? The lexicon is the area of language where idiosyncrasy abounds and one should not be surprised, for instance, at the existence of time drift as in cockpit: 1587 “a pit for cockfights” > 19th cent, “a depressed area of the deck near the stern of a sailing boat from which she is steered” > 1914 “a space occupied by the pilot in a plane or 1935 racing car”.20 Beside polysemy, changes of denotatum or homonymous re-creations contribute to blurring the picture as with airline: 1813 (USA) “a straight line between two places”, 1914 “an aircraft transport service”. It is difficult to imagine how such units could be dealt with if they were not listed in the lexicon. In the case of novel endocentric compounds, however, the hearer obviously has to combine the semantic representations of the components with pragmatic information to deduce the nature of the modification link. Psycholinguistic research in this area is represented by concept combination studies, with several models of interpretation strategies. In one, the representation of the compound’s head includes a number of slots and their possible or frequent fillers; in the combination process, items from the representation of the modifier fill some of the slots (see Wisniewski 1996 for a review). To Gagné and Spalding (2006), relations are more important than the representations of the components in the formation of the combined concept, and the selection of a N1-N2 relation can be expected to depend on which modification relation Ν1 was previously used with. It is the relation that adds features absent from the representation of N2 to the combined concept. This in turn poses the question of the cognitive nature and inventory of the relations: for instance, do they exist independently from compounding or are they emergent phenomena?

  • 21 See for instance Kövecses and Radden (1998).
  • 22 Admittedly not an [N1N2]N compound but a genitival one.
  • 23 This is not due to the N1s in duckbill and cranesbill being bird names, since parrotbill does denot (...)

47As to novel secondary exocentrics, their comprehension certainly benefits from analogy with pre-existing tropes (Benczes 2006: 2), but this strategy would not apply to cases where the tropes themselves are novel. Research, particularly taxonomy-oriented work,21 on metaphor and metonymy can contribute by providing probabilistic paths: for instance, when test A fails and N2 is head, what are the chances of the modification relation being of the contents-container category, as in airhead, and referring to a human? When N2 is bill, does [N1N2]N denote a bird as in spoonbill or hornbill? (Duckbill is synonymous with platypus, however, and cranesbill22 denotes a plant!23). Benczes (2006) applies to exocentric compounds cognitivist representations of conceptual metaphor and metonymy in combination with blending theory (Fauconnier and Turner 2002). As the author herself acknowledges (p. 140), one problem with this approach, however, in particular with respect to the diagrams of blends, is the subjectivity of some of the analyses. Other research is taking place in a more formal framework. One direction consists in the application of the word representations, in particular the qualia, of Pustejovski’s (1995) Generative lexicon. Bassac and Bouillon (2001) investigated the Telic in French and Turkish compounds involving different qualia and Bassac (2006) presents schemata corresponding to the derivation of the meanings of examples such as fruit juice and wine glass. Interestingly, Bassac mentions that it was found necessary to complexify the Telic qualia and one suspects that more adjustments may be in order when less simple examples are tackled.

48Jackendoff (forthcoming) combines “basic functions” like LOCtemp (X,Y) and cocomposition from the representations of the components and manages to account for rather complex cases like piano bench (“a bench on which one sits while playing the piano”), and even more complexity can probably be dealt with at the cost of an increase in the intricacy of the schemata. He presents schemata for secondary exocentrics like canvasback (the name of a duck) or birdbrain using a function, similar. The schema for canvasback includes the representation of duck, but what isn’t clear is how this kind of information is accessed when the context does not point to an obvious referent. The author explicitly asks the question of the number of existing basic functions, and research on this point clearly isn’t over. Among the advantages of Jackendoff’s system is that he doesn’t shy away from exocentrics which, as Benczes (2005) notes, have been largely avoided so far by theorizing approaches.

49And, try as we might, some compounds that condense extensive stories by an accumulation of tropes like soap opera, light year, couch potato, chorus line or agony aunt, with the help of either lexicalization or pragmatic information, will probably resist analysis for a long time, and research on the semantics of compounding obviously has a long future.

Top of page

Bibliography

Adams, V. (1973). An Introduction to Modern English Word Formation, London: Longman.

Allen, M.R. (1978). Morphological Investigations, Ph.D. dissertation, University of Connecticut, Storrs, Ct.

American Heritage College Dictionary, 3rd edition (1997), Jost, D.A., Ed., Boston: Houghton Mifflin.

Arnaud, P.J.L. (2003). Les Composés timbre-poste, Lyon: Presses Universitaires de Lyon.

Barcelona, Α., Ed., (2000). Metaphor and Metonymy at the Crossroads: A Cognitive Perspective, Berlin: Mouton-de Gruyter.

Barcelona, A. (2003). “Clarifying and applying the notions of metaphor and metonymy within cognitive linguistics: An update”, in Dirven, R., Pörings, R., Eds., Metaphor and Metonymy in Comparison and Contrast, Berlin: Mouton-de Gruyter, pp. 207-277.

Bassac, C. (2006). “A compositional treatment for English compounds”, Research in Language, 4: 133-153.

Bassac, C., Bouillon, P. (2001). “The telic relationship in French and in Turkish compounds”, paper presented at the Second International Workshop on Generative Approaches to the Lexicon, ETI Geneva, University of Geneva.

Bauer, L. (1998). “When is a sequence of two nouns a compound in English?”, English Language and Linguistics, 2: 1, pp. 65-86.

Bauer, L. (2006). “Compounds and minor word-formation types”, in Aarts, B., McMahon, Α., Eds., The Handbook of English Linguistics, Oxford: Blackwell.

Bauer, L. Huddleston, R., (2002). “Lexical word-formation”, in Huddleston, R., Pullum, G.K., Eds., The Cambridge Grammar of the English Language, Cambridge: C.U.P., pp. 1621-1721.

Bauer, L., Renouf, A. (2001). “A corpus-based study of compounding in English”, Journal of English Linguistics, 29: 2, pp. 101-123.

Benczes, R. (2005). “Metaphor-and metonymy-based compounds in English: A cognitive linguistic approach”, Acta Linguistica Hungarica, 52: 2-3, pp. 173-198.

Benczes, R. (2006). Creative Compounding in English, Amsterdam: Benjamins.

Bisetto, Α., Scalise, S., (2005). “The classification of compounds”, Lingue e Linguaggio, 4: 2, pp. 319-332.

Bloomfield, L. (1933). Language, New York: Holt.

Brekle, H.E. (1970). Generative Satzsemantik und transformationelle Syntax im System der englischen Nominalkomposition, München: Fink.

COD11 = Concise Oxford English Dictionary, 11th edition (2004). Soanes, C., Stevenson, Α., Eds., Oxford: O.U.P.

Fauconnier, G., Turner, M. (2002). The Way we Think: Conceptual Blending and the Mind’s Hidden Complexities, New York: Basic Books.

Gagné, C.L., Spalding, T.L. (2006). “Conceptual combination: Implications for the mental lexicon”, in Libben, G., Jarema, G., Eds., The Representation and Processing of Compound Words, Oxford: O.U.P., pp. 145-168.

Geeraerts, D. (2003). “The interaction of metaphor and metonymy in composite expressions”, in Dirven, R., Pörings, R., Eds., Metaphor and Metonymy in Comparison and Contrast, Berlin: Mouton-de Gruyter, pp. 435-465.

Goossens, L. (2003). “Metaphtonymy: The interaction of metaphor and metonymy in expressions for linguistic action”, in Dirven, R., Pörings, R., Eds., Metaphor and Metonymy in Comparison and Contrast, Berlin: Mouton-de Gruyter, pp. 349-377.

Hansen, B., Hansen, K., Neubert, Α., Schentke, M. (1982), Englische Lexikologie: Einführung in Wortbildung und lexikalische Semantik, Leipzig: Verlag Enzyklopädie.

Hatcher, A.G. (1960). “An introduction to the analysis of English noun compounds”, Word, 16: 356-373.

Jackendoff, R. (forthcoming). “The Ecology of English Noun-Noun Compounds”, in Lieber, R., Štekauer, P., Eds., Handbook of Compounding, Oxford: O.U.P.

Jespersen, O. (1942). A Modern English Grammar on Historical Principles, London: Allen & Unwin.

Kövecses, Z., Radden, G. (1998). “Metonymy: developing a cognitive linguistic view”, Cognitive Linguistics, 9: 37-77.

Kuiper, K. (1999). “Compounding by adjunction and its empirical consequences”, Language Sciences, 21: 407-422.

Lees, R.H. (1970). “Problems in the grammatical analysis of English nominal compounds”, in Bierwisch, M., Heidolph, K.E., Eds., Progress in Linguistics, The Hague: Mouton, pp. 174-186.

Levi, J.N. (1978). The Syntax and Semantics of Complex Nominals, New York: Academic Press.

Maetzner, E. (1860/1873). Englische Grammatik. s.l., B. Weidmann [1st ed., I860].

Noordegraaf, J. (1989). “Front the history of the term ‘endocentri’”, Historiographia Linguistica, 16: 211-215.

OED = Oxford English Dictionary, on-line edition, Oxford: O.U.P.

Pustejovsky, J. (1995). The Generative Lexicon, Cambridge (Massachussetts): M.I.T. Press.

Renner, V. (2006). Les Composés coordinatifs en anglais contemporain, Doctoral dissertation, Université Lumière, Lyon, [available online].

Renner, V. (forthcoming). “On the Semantics of English Coordinate Compounds”, English Studies.

Selkirk, E.O. (1982). The Syntax of Words, Cambridge (Massachussetts): M.I.T. Press.

Søgaard, A. (2004). “A Compound Matrix”, in Müller, S., Ed., Proceedings of the HPSG04 Conference, Stanford: CSLI Publications, pp. 443-455.

Warren, B. (1978). Semantic Patterns of Noun-Noun Compounds, Göteborg: Acta Universitatis Gothoburgensis.

Warren, B. (1992). Sense Developments, Stockholm: Almkvist & Wiksell.

Wisniewski, E.J. (1996). “Construal and similarity in conceptual combination”, Journal of Memory and Language, 35: 434-453.

Top of page

Notes

1 A variant of this test,
(a) [N
1N2]N is a kind of N2
is actually a test of hyponymy and may not return the same results:
a police car is a car ~? a police car is a kind of car.

2 Some of the examples are restricted to the British or North-American dialect zones; tractor-trailer = articulated lorry.

3 See Noordegraaf (1989) for a history of endocentric before Bloomfield.

4 Or, if a clear hyperonymic lexeme does not exist (which is the case for singer and songwriter), they are at the same level of a conceptual hierarchy.

5 Bauer (1998, 2006) discusses the limitations of using stress as a criterion of compounding.

6 Dvandva, as well as bahuvrihi (to be introduced further on), are terms borrowed from Sanskrit linguistics.

7 The American Heritage College Dictionary, 3rd ed., has bluefin tuna; COD11, however, has an entry bluefin.

8 Airhead obviously never denoted a head in the first place and it is actually a convenient fiction to consider that the trope operates after the N1 to N2 modification.

9 Readers may accuse me of having selected a first set of endocentrics that test positive on C, while another set and a different test would have produced a different view. A counter-argument is that analogy is a well identified independent cognitive principle that stands out in clear contrast with the other relations in compound modification.

10 In the form of 13 underlying generalized verbs plus 6 relations that cannot be generated thus.

11 I find it preferable to restrict the term synthetic to [NVer]N and [NVing]N units like circuit breaker and sleep walking. I also use the term modification even when the restriction on N2 is provided by an argument, as this does not complicate the issue.

12 There is also a growing minority of non-analogical [N1N2]N units like cas sujet, code barre, capital risque, bière pression (Arnaud 2003).

13 For a similar analysis on guard dog, see Jackendoff (forthcoming).

14 Some of the material in this section was presented at the Journée d’études sur les composés, University of Arras (France), 24 March 2006.

15 In an adequate context bottom line could have literal reference, but this sequence is probably not lexicalized (at least there is no mention of it in COD 11).

16 The difference with Leatherstocking, the proper name of the eponymous character of a series of novels by J. Fenimore Cooper, is that in this case leather and stocking are literal before the operation of the overall metonymy.

17 We could probably add rattlesnake to this list, but my informants were unable to choose between a rattlesnake is a snake that has a rattle ~ a rattlesnake is a snake that rattles.

18 When N1 is [human], the tendency is for semantically similar compounds to be genitival as in busman’s holiday or director’s chair.

19 Have” is a label for a non-merological abstract relation which could equally well be labelled “Of”:
HAVE (bear, hug) = OF (hug, bear).
The
HAVE relation dominates the merological one.

20 OED and World Wide Words website.

21 See for instance Kövecses and Radden (1998).

22 Admittedly not an [N1N2]N compound but a genitival one.

23 This is not due to the N1s in duckbill and cranesbill being bird names, since parrotbill does denote a bird.

Top of page

References

Bibliographical reference

Pierre J.L. Arnaud, « Semantic Complexity in English [NN]n Compounds », Anglophonia/Sigma, 12 (24) | 2008, 7-21.

Electronic reference

Pierre J.L. Arnaud, « Semantic Complexity in English [NN]n Compounds », Anglophonia/Sigma [Online], 12 (24) | 2008, Online since 13 December 2016, connection on 28 June 2017. URL : http://anglophonia.revues.org/962 ; DOI : 10.4000/anglophonia.962

Top of page

About the author

Pierre J.L. Arnaud

CRTT, Université Lumière-Lyon 2

Top of page

Copyright

Licence Creative Commons
Anglophonia – French Journal of English Linguistics est mis à disposition selon les termes de la licence Creative Commons Attribution - Pas d'Utilisation Commerciale - Pas de Modification 4.0 International.

Top of page
  • Logo Presses universitaires du Midi
  • Revues.org