You are on page 1of 9

1 Christopher Desurmont1

Postnominal adjectives2

The introduction develops what is said about postpositive adjectives in Greenbaum and Quirk 1990 (taken as a typical example of what is said in textbooks). The rest of this paper investigates the status of postnominal adjectives more closely. It has often been suggested that postnominal adjectival phrases (APs) can or should be regarded as "reduced relative clauses"; I begin with cases that can indeed apparently be considered as such. Then I try to show that other postpositive adjectives (or APs) cannot be analyzed as RRCs. This paper ends with a discussion on postnominal participle-looking ing words that are examined in the light of Kibort 2005's claim that they should be treated as "category neutral"; hence, that their status (more or less adjectival/verbal) depends on position and distribution.
Introduction

Greenbaum and Quirk 1990: 136-137 (henceforth: G & Q) give the following example of a "postpositive adjective": something useful, saying that "a postpositive adjective can usually be regarded as a reduced relative clause: "something that is useful". Normally, one cannot in English place an adjective (here qualifying or evaluative, hence speakeroriented) in prenominal position in front of an indefinite pronoun: *a useful something3. G & Q then talk of "institutionalized expressions" in which the adjective can only be postpositive: president elect, heir apparent, attorney general, body politic, to name a few. Other examples would be princess royal, moment supreme. G & Q do not indicate how these adjectives are to be analyzed.

Universit Lille 3. Communication la journe de l'agrgation sur les adjectifs (ALAES), 3 fvrier 2007, Paris 4. 3 Abney (1987), reasoning from a generative perspective, examining the DP (determiner phrase) something wild proposes a derivational account starting from the DP some wild thing and involving head movement of N (head of NP) to D (head of DP). As Svenonius (1994) points out, this transformational view doesn't explain why postnominal APs are non-iterable; and to this I would add: ... knowing that conversely, prenominal APs can be stacked (there can be more than one) as in some wild extravagant thing where there are two prenominal adjectives. Indeed, you can have one postnominal AP but not more than one: *something wild extravagant3. For further reading on this issue, see Larson & Marui 2004 for instance.
2

Looking at what Giegerich (2006) says in the following passage for example, these constructions would be analyzed as complex lexical units of a sort, which means that the adjective does not have the status of an adjunct: "It is a probably uncontroversial assumption about formal grammar that all constructions originating in the syntax are essentially transparent and the outcomes of fully productive processes; non-transparent or non-productive constructions (red herring, court martial) are by definition lexical." A "transparent" expression is one in which each lexical item (in this case: A and N in the A/N relation) retains its normal meaning even though one of the two modifies the other (A modifies N, or the referent of N). For example, in the NP (or DP) [atomic energy], atomic is a denominal "relational" adjective4 meaning "from the atom", and the head noun energy is used in its usual sense, whereas in [attorney general], the adjective general does not have the same ordinary meaning as (for instance) in a general mess, or even in a general assembly. Summing up "the common pre-theoretic notions of how adjuncts differ from complements", Dowty (2006) writes : "(...), the same adjunct combined with different heads affects their meaning in the "same" way semantically (e.g. walk slowly vs. write slowly)." According to this criterion, the postpositive adjective general in attorney general would not be regarded as an adjunct. Whatever the exact nature of these constructions, the postnominal adjective cannot be turned into a relative clause: *a court which is martial, *a president who is elect, *an attorney who is general. G & Q go on to say that "adjectives ending in able and ible can be postpositive, as well as attributive (...)". Thus one can say the best use possible or the best possible use. The authors here talk of "a discontinuous adjective". I shall return to these superlative constructions in part 2. G & Q (repeating what Bolinger 1967 said) then indicate that some adjectives can be both postnominal or attributive, stars visible or visible stars. Postnominal APs "convey the implication that what they are denoting has only a temporary application. Thus, the stars visible refers to stars that are visible at a time specified or implied, while the visible stars more aptly refers to a category of stars that can (at appropriate times) be seen." If the property expressed by the adjective (A) is a permanent and defining characteristic of the referent of the head noun (N), this means that A is a property that characterizes the object prior to the moment of speech; and this is encoded in the syntax by placing A in prenominal rather than postnominal position. G & Q then give an example of a property that is clearly non-lasting (transient) because linked to a particular moment or situation, and hence expressed postpositively (post-nominally): The soldiers present were his supporters.
1. Postnominal APs that admit a RRC reading

We recall that Greenbaum. & Quirk 1990:136 suggest that postnominal APs can be considered as "reduced relative clauses": something useful = something that is useful. Others think likewise, for instance, Svenonius (1994): "I will not discuss predicative adjectives used as postmodifiers, assuming them to be some sort of reduced relative, right-adjoined to NP or DP like PPs and relative clauses and subject to various restrictions (they are necessarily interpreted as stage-level5; cf. The visisible stars versus the stars visible)".

As such it it also an "absolute" or "intersective" adjective; i.e. one whose meaning is not "relative" in the sense that it does not vary according to the nouns it modifies. 5 The term "stage-level" refers to "stage-level predicates", which are predicates valid only in a specific non-lasting situation.

The AP visible can apparently be seen as a RRC (that are visible) if Svenonius is right when he qualifies the (typical?) postpositive adjective as being "predicative", since a "predicative adjective" is by definition one placed to the right of a copular verb like BE where it is "predicated of" the subject NP (or DP); and this copular verb appears in the relative clause. A RRC reading does look equally plausible in the following example as well (mine): A girl [AP afraid of insects] won't want to sleep outside. A girl who is afraid of insects etc. Here, the AP is adjoined to the NP girl and there is a cause and effect relationship between "being afraid of insects" and "not wanting to sleep outside". Such a relationship means that the constituent that is being predicated of is not *[girl], but [girl afraid of insects]. In other words, the postAP has to have a restrictive reading for the sentence to have the intended meaning, i.e., the only plausible meaning. Like in ellipsis, according to the traditional definition of ellipsis as expressed in Fontanier (1830) 1. the missing words are unequivocally retrievable (namely who is), and 2. the filling-in process yields a constituent that has the same meaning. The property denoted by a postnominal adjectival phrase (henceforth: postAP) can be at the same time transient and restrictive, while a prenominal adjectival phrase (henceforth: preAP) will denote a property that is often ambiguous between a restrictive and a nonrestrictive reading (Bolinger 1967). Here's an example from Larson & Marui (2004): (a) Every blessed person was healed. 'All the people were healed.' 'All the people that were blessed were healed.' (b) Every person blessed was healed. # 'All the people were healed.' 'All the people that were blessed were healed.' Sentence (a) with a preAP is ambiguous, because in prenominal position, the adjective can be read parenthetically (= non-restrictive reading), or non-parenthetically (= restrictive reading). Sentence (b) is not ambiguous since in the postnominal position, the adjective receives a restrictive reading only. In the words of the authors: "(...) the prenominal adjective can have a parenthetic character, but the postnominal adjective cannot." So a postAP (as in sentence b) cannot be interpreted with a non restrictive reading. We now observe the possibility of placing postnominal blessed in a relative clause: (c) Every person who was blessed was healed.

Two observations must be made: 1. that the most obvious interpretation of this relative clause is a defining one; so this clause is restrictive; and 2. that with this restrictive reading of the relative clause, sentence (c) has the same meaning as sentence (b), with blessed in postnominal position, but not the same meaning as sentence (a) where blessed is prenominal since (a) has two possible meanings (i.e., is ambiguous). If every person blessed means every person who is blessed with a restrictive reading of the relative clause, this would explain for instance why in the next sentence (mine), the elliptical VP at the end can only mean "(...) that they were healed", not "that they were blessed". (d) Every person blessed was healed, though not everyone would agree that they really were.

Indeed, if the postAP can only admit a restrictive reading (a non parenthetical one), then interpreting the elliptical VP as meaning not everyone would agree that they were really blessed would be a contradiction. Let's now take another example: The person [AP responsible] is asked to speak up. We can immediately check that this AP can be turned into a (defining) relative clause without changing the meaning of the sentence: The person who is responsible is asked to speak up. With a definite DP (the determiner here is the definite article the), and in this particular context (we're looking for the person who is responsible for the misdeed), the relative clause can hardly receive a non restrictive reading; i.e., it can only be a defining relative clause.
2. Postpositive APs not analyzed as RRCs.

I now look at postpositive APs for which arguments against a RRC reading can be found.
A Superlative DPs

Postpositive APs are often found associated with prenominal superlatives: The best hotel available The worst conditions imaginable The cheapest trip possible First, we see that *the cheapest trip that is possible, is not acceptable. To explain this, one could invoke the close link between postnominal possible and the prenominal superlative adjective. For Schwartz, possible is exclusively linked to the superlative morpheme, and superlatifs NPs are not derived structures; more specifically, structures such as these do not arise from "promotion-and-ellipsis" as Larson (2000) claims. Schwartz (2005) says that in the largest present possible [for anyone to carry], "largest possible [is] a constituent, parsed as large [est possible]"6. If this is so, the RRC reading would make no sense.
B Verbless clause-like subjects (small clause)

Now if we look at the following sentence (found in Bolinger 1952): (a) Mary beautiful is something hard to imagine.

These are his arguments: 1) Comparing Mary bought the largest present possible [for anyone to carry],et Mary bought the largest possible present [for anyone to carry], he observes that the meaning has changed; in particular, sentence n 2 (where possible is now placed just right of the prenominal superlative) admits an interpretation of the type "purpose clause" (a present heavy enough so that no one will be able to carry it). Sentence n 1 cannot have this meaning. To avoid this "purpose-like" reading, ellipsis of the infinitive clause is necessary, and this is not ellipsis's traditional role. 2) Possible cannot be placed to the right of an analytical superlative:*Mary bought the most expensive possible present. 3) Possible) placed tothe right of a compound superlative adjective will be interpred not in relation to this superlative, but as an epistemic attributive adjective bearing upon the head noun. Thus, in They hired the best-known possible philosopher, possible is linked to the noun, not to the superlative. Schwartz concludes saying: "This suggests that prenominal possible never involves promotion and ellipsis in the first place. An alternative account analyses largest possible as a constituent, parsed as large [est possible]".

... we see that the AP beautiful cannot be a relative clause of any sort: (b) *Mary who is beautiful is something hard to imagine. Yet what is hard to imagine is indeed that Mary should ever manage to be beautiful. This suggests that the subject in (a) might be a "small clause". Sentence (c) is not very good, but it sounds better than (b): (c) ? Mary being beautiful is something hard to imagine. This sentence seems to show that the subject is some sort of clause in which the copular verb has been omitted. There seem to be referential constraints of some sort (that would require further investigation) that allow or disallow the use of this structure. Compare: (a) (a') (d) (d') Mary beautiful is something hard to imagine. *Mary beautiful today is good news. Mary fed up with herself is not hard to imagine. ??Mary fed up with herself is a pity.

The structure seems to be fully licensed only if the predicative relation of the small clause is shown through the predication not to be validated. This relation can be presented as plausible as in Mary fed up is not hard to imagine, but it cannot be shown through the predication to have already been validated.
C Stage-level predicates in presentational "there" sentences

Compare: (a) There are firemen available ("presentational" reading) (b) There are firemen who are available. ("existential" reading) Sentence (a) has the adjective available in postnominal position, and can only have a presentational reading; i.e."the firemen are there somewhere now as I speak, and they are available for intervention". This sentence cannot receive an existential interpretation of the type There exist firemen who are available people, which has a contrastive implication that sentence (a) does not have. Now if one looks at sentence (b), one can check that the most obvious reading is not the presentational one, but the existential one; i.e. a reading 1. that defines a sub-class of firemen, 2. has a contrastive value (i.e., suggesting that there exists another sub-class of firemen who are (possibly) never available), and 3. is not situationally anchored, contrary to sentence (a) that we are examining. The conclusion is that sentence (a)'s postAP is not a RRC. If one wants to reach a more general conclusion, then I think one could call upon the distinction between three kinds of predicates (Carlson 1977, 1980): stage-level, individual-level and kind-level predicates. The first type of predicate is only valid for a time, in a specific context or situation; the second is true of the individual presumably for a lifetime; and the third is true only for a kind of N, which means that the subject NP or DP is generally plural (Humans are widespread, not *a human is widespread). There are firemen available is clearly a stage-level predicate, like for instance John is hungry, or Bill is being a nuisance, as opposed to Bill is a nice guy which is an individual level predicate.

This brings me to the conclusion that a stage-level postpositive AP in a presentational sentence (There is/are ...) cannot be a (defining) RRC.
D Deverbal ed adjectives (and complementation)

(a) *The money is deposited (b) *He has money which is deposited in the bank (c) He has money which has been deposited in the bank. (d) He has money deposited in the bank. (e) *He has money deposited in the bank by his parents. (f) ??He has money deposited in the bank for him (g) He has had money deposited in the bank for him by his parents. This is how Dirven 1999: 61-62 (basing his remarks on Bolinger 1967) explains the non grammaticality of sentence (a): "There is no constraint on the post-posed participle: money deposited or withdrawn are both possible .... In pre-position only the participle denoting a relevant, lasting-state is acceptable; deposited money, but not *withdrawn money. But such "lasting-state" participles like deposited cannot make it for a bepredication; "the money is deposited. Such a predicative use is only possible when the past participle has become fully equivalent to a freely attributive adjective". If sentence (a) is not acceptable, then (b) with a relative clause is unacceptable as well for the same reason. In this context, our discussion on RRCs comes to a stop. However, I'd like to consider another question: Why is sentence (d) acceptable, but neither (e) nor (f), seeing that there is no be-predication here? Deposited is adjectival in (d); and this is why there can be no agent-complement: sentence (e) is therefore not acceptable with its by-complement: by his parents. But why should (f) be unfelicitous, seeing that there is no by-complement? (d) He has money deposited in the bank. (f) ??He has money deposited in the bank for him [deposited in the bank] is a complement to the noun money7, i.e. an "object complement", and the verb HAS "localizes" [money deposited in the bank] with respect to the subject He. Considering the meaning of the two constituents on either side of the verb, and a number of other traits such as the fact that He refers to a human being, HAS is interpreted in terms of "possession", or possibly "location" (of a sort). Now, adding the prepositional phrase [for him] has the effect of giving the subject NP the role of "beneficiary" or "goal". If this sentence sounds a little odd, the explanation might be that the two semantic roles (possessor and beneficiary, or possessor and goal) for the one subject NP are (somewhat) incompatible. To put it another way, without the PP [for him], the event described is static (a result); but with the addition of this constituent, the event described is at the same time static and dynamic; and the questionable grammaticality of this sentence would indicate that one cannot express the two points of view at the same time.
3. The status of participle-looking ing words.

See Radford 1988: 175-196 for an explanation of the difference between "adjuncts" and "complements". The traditional criteria are summed up also (for instance) in Dowty, The Dual Analysis of Adjuncts/Complements in Categorial Grammar.

I shall now be arguing that the possibility of turning a postpositive participle-looking ing word into a relative clause proves nothing regarding the status of this word, verbal or adjectival, and shall plead in favour of Kibort's 2005 "category-neutral" description of this type of word, but not without some reservation however. Compare8: (a) (b) (c) an interesting book an interesting child a smiling/crying child

Are these ing words fully adjectival, or are they more verbal-like, i.e., participles? Bresnan 1995 proposed a battery of tests to distinguish adjectives from participles. One of them says that a present-participle-looking word must be an adjective if it can be attributive, i.e. prenominal: A N vs. *V N. Hence, smiling and interesting would both be adjectives. Laczk (2001) says that some participle-looking words always require a complement, and therefore cannot be prenominal in any case.; so this test cannot apply to them. Now compare: (d) a very interesting child (e) *a very smiling child. (f) a very big smile / a bigger smile than that The non-grammaticality of (e) cannot be due to nongradability, since (f) is o.k., but it may well be due to the adjective's "participle" nature. Smiling as opposed to interesting always implies agentivity on the part of the referent of N. It denotes an action that can be done to a greater or lesser degree, whereas interesting denotes a property. Let us assume with Montague 1970: 211, (building upon proposals by Kamp and Parsons) that "the denotation of an adjective phrase is always a function from properties to properties". Smiling denotes an activity, and only by way of consequence can a property be inferred: for instance, a smiling child might be a happy child, and happy is clearly a property. A distinction may be needed between -ing words in the NP domain denoting properties directly (therefore directly inherent or ascriptive), and ing words primarily denoting an activity (and from which a property may possibly be inferred). Now let us compare: (g) a book interesting to discuss (h) a child smiling at you. One of Bresnan's tests applies only to participle-looking words that are derived from transitive verbs. If they take a direct complement, they cannot be adjectives, they are participles: a girl watching you, for instance, and indeed, watching cannot be attributive (*a watching girl). The verb smile is intransitive (or indirectly transitive), so this test cannot be applied. Another test says that only an adjective can head concessional relative phrases: However interesting the book is, I won't have time to read it. *However smiling the child is, he won't play with the others. However much the child smiles, etc.
8

From a "role-based" point of view, there are three situations here. In (c), clearly the referent of the head noun is doing something, and is agentive. In (b), the AP expresses a judgment either about a child's personality, or about what the child is presently doing; so N could be agentive, but not necessarily. In (a), the book cannot actually be "doing" anything, so it isn't be agentive.

So this test indicates that smiling is not an adjective, that it is a participle; and this is the opposite conclusion to what the attributive test claims ( A N vs. *V N). To check the status of smiling, turning postpositive segments into relative clauses is inconclusive, because it works just as well with interesting which is clearly adjectival as with smiling or crying whose status we are investigating. (i) a book that is interesting to discuss. (j) a child who is smiling at you. This transformation does not prove that the participle-looking word placed in postnominal position has verbal rather than adjectival status, for we know that interesting is essentially adjectival. So the mere syntactic possibility of turning a postnominal ing word into a relative clause doesn't help to demonstrate the status of this word. In (j), is smiling is a present continuous (with be-ing as an aspectual marker), and in (i), interesting to discuss is a predicative AP. This difference in status should (somewhat) block coordination, and apparently it does: ??A smiling and interesting child ??An interesting and smiling child a smiling and convincing look a persuasive and smiling attitude The difference in status between two participle-looking words (smiling and interesting ) does not exclude the possibility that one and the same -ing word can have two different status according to which position it occupies. Smiling would thus be more adjective-like in prenominal position, but verbal-like in postnominal position. The adjective in A N would be obtained through some sort of participle-adjective conversion rule (Bresnan 1978:8-9, 1982:29, 2001:31). Anna Kibort (2005) does not entirely question Bresnan's conversion rule, but claims that "the input to the Conversion Rule is a verb and the output is a deverbal form (a participle) which is category-neutral between a verb and an adjective (i.e. it is both a verb and an adjective at the same time)." If Kibort is right, one might then suggest that what we have is indeed a "category neutral" deverbal ing word whose status is more or less verbal/more or less adjectival according to its position within or without an NP domain. Such a position would account for the following facts or observations: 1. that smiling can be attributive, therefore adjectival-like (a smiling child); 2. that its dual nature is maintained in every position, which prevents it from being prototypically adjectival even in prenominal position (*a very smiling child); its limited adjectival status being revealed by the however test; 3. that it can head a VP, in which it clearly retains its original verbal character: the child is [VP smiling at you].

Bibliography
Bolinger, D., (1967). Adjectives in English: Attribution And Predication, Lingua 18. NorthHolland Publishing Co., Amsterdam. 1-34. Bresnan, J., (1978). A realistic transformational grammar. In Halle, M., Bresnan, J., & Miller, G.A. (eds.), Linguistic Theory and Psychological Reality. Cambridge, MA: The MIT Press. 1-59. Bresnan, J., (1982). The passive in lexical theory. In Bresnan, J.(eds.), The Mental Representation of Grammatical Relations. Cambridge, MA: The MIT Press. 3-86. Bresnan, J., (2001). Lexical-Functional Syntax. Oxford:Blackwell. Bresnan, J. (1995). Lexicality and argument structure, in: Proceedings of the Paris Syntax and Semantics Conference. Carlson, G., N., (1977). A unified analysis of the English bare plural. Linguistics and philosophy, 1:3, 413-458. Carlson, G., N., (1980). Reference to Kinds in English. New York: Garland Publishing. Dirven, R., (1999). The Cognitive Motivation For Adjective Sequences in Attribution, Journal of English Studies, I. 57-67. Dowty, D. The Dual Analysis of Adjuncts/Complements in Categorial Grammar. http://www.ling.ohio-state.edu/~dowty/papers/degruyter.8 11.pdf Fontanier, P. 1830, (1968). Les Figures du Discours. Flammarion. Giegerich, H. (2006). Attribution in English and the distinction between phrases and compounds. http://www.englang.ec.ac.uk/people/attributioninenglish.pdf Greenbaum, S., & Quirk, R. (1990). A students grammar of the English language, Longman. Kibort, A, (2005) The Ins and Outs of the Participle-Adjective Conversion Rule. Proceedings of the LFG 05 Conference, University of Bergen, Miriam Butt and tracy Holloway King (Editors), CSLI Publications. http://csli-publications.stanford.edu Laczk, T., (2001). Another Look at Participles and Adjectives in the English DP. Proceedings of the LFG01 Conference, University of Hong Kong, Miriam Butt and Tracy Holloway King (Editors), CSLI Publications. http://csli-publications.stanford.edu Larson, R., (2000). ACD in AP?, paper presented at the 19th West Coast Conference of Formal Linguistics, Los Angeles. http://semlab5.sbs.sunysb.edu/~rlarson/larson-papers.html Larson, R., & Marui, F. (2004). On indefinite pronoun structures with APs: Reply to Kishimoto. Linguistic Inquiry 35 (2): 268. http://www.msu.edu/~morzycki/papers/nonrestrict.pdf Montague, R., (1970). English as a Formal Language. In Linguaggi nella Societ e nella Tecnica, ed. Bruno Visentini et al.. Milan: Edizioni di Comunit. 189-224. Quirk, R., Greenbaum, S., Leech, G., Svartvik, J., (1985). A Comprehensive Grammar of the English Language. London: Longman. Radford, A, (1988). Transformational Grammar, A First Course. C.U.P. Schwartz, B., (2005). Modal superlatives. Proceedings of SALT 15 (Semantics and Linguistic Theory), University of California Los Angeles (UCLA). http://www.linguistics.ucla.edu/SALTFiles/SchwartzSalt15.pdf Svenonius, P. (1994). The Structural Location of the Attributive Adjective, in: E.Duncan, D.Farkas, and P.Spaelti (eds.), The proceedings of the Twelth West Coast Conference on Formal Linguistics, CSLI, Stanford (Ca.). 439-454.

You might also like