Uroš Martinčič	DOl: 10.4312/elope.11.2.21-35
Slovenia
Encasing the Absolutes
Summary
he paper explores the issue of structure and case in English absolute constructions, whose subjects are deduced by several descriptive grammars as being in the nominative case due to its supposed neutrality in terms of register. his deduction is countered by systematic accounts presented within the framework of the Minimalist Program which relate the case of absolute constructions to specific grammatical factors. Each proposal is shown as an attempt of analysing absolute constructions as basic predication structures, either full clauses or small clauses. I argue in favour of the small clause approach due to its minimal reliance on transformations and unique stipulations. Furthermore, I propose that small clauses project a singular category, and show that the use of two cases in English absolute constructions can be accounted for if they are analysed as depictive phrases, possibly selected by prepositions. he case of the subject in absolutes is shown to be a result of syntactic and non-syntactic factors. I thus argue in accordance with Minimalist goals that syntactic case does not exist, attributing its role in absolutes to other mechanisms.
Key words: absolute constructions, case, Minimalist syntax, small clauses
O zgradbi in sklonu v absolutnih zgradbah
Povzetek
Članek obravnava vprašanje zgradbe in sklona v angleških absolutnih zgradbah, ki po sklepanju mnogih opisnih slovnic vsebujejo osebek v imenovalniku na podlagi domnevne registrske nevtralnosti tega sklona. Nasproti temu sklepu so navedeni sistemski pristopi, ki so predstavljeni v okviru minimalistične teorije jezika in ki povezujejo sklon v absolutnih zgradbah z ustreznimi slovničnimi dejavniki. Vsak pristop je prikazan kot poskus analize absolutnih zgradb kot osnovnih predikacijskih zgradb, bodisi celih bodisi malih stavkov. Slednji pristop zagovarjam kot ustreznejšega, saj se v manjši meri zanaša na pretvorbe in edinstvene predpostavke. Predlagam tudi, da mali stavki razvijejo enotno kategorijsko zgradbo, in prikažem, da je rabo dveh sklonov v angleških absolutnih zgradbah mogoče razložiti, če so absolutne zgradbe depiktivne besedne zveze, ki lahko nastopijo kot dopolnila predlogov. Sklon osebka v absolutnih zgradbah je prikazan kot posledica skladenjskih in neskladenjskih dejavnikov. Temu ustrezno v skladu z minimalističnimi prizadevanji zagovarjam hipotezo, da skladenjski sklon ne obstaja, in njegovo vlogo v absolutnih zgradbah pripišem drugim mehanizmom.
Ključne besede: absolutne zgradbe, sklon, minimalistična skladnja, mali stavki
Encasing the Absolutes 1. Introduction
Absolute constructions present an intriguing puzzle for theories of grammar, because they typically lack any discernible connection to the matrix clause. hey are also exceptional clauses in that they contain an overt subject despite their non-finiteness. Furthermore, English absolute clauses may contain a considerable variety of predicates. As shown by the following examples, these range from nominal, adjectival, and prepositional categories to the verbal one, which may host any of the non-finite verb forms (the infinitival being the least frequent):
(1)	a. 73 people have been drowned in the area, most of them children.
(Quirk et al. 1985, 996)
b.	73 people have been drowned in the area, most of them still young.
c.	73 people have been drowned in the area, most of them still in their teens.
(2)	a. he discussion completed, the chairman adjourned the meeting for half an hour.
(Quirk et al. 1985, 993)
b.	Her aunt having left the room, I asked Ann for some personal help. (loc. cit.)
c.	We shall assemble at ten forty-five, the procession to start at precisely eleven.
(Visser 1972, 1056)
he external structure of absolute clauses is also diverse. his is seen in the fact that absolute constructions can be introduced by the preposition with or its negative equivalent without. Stump (1985) refers to instances of absolutes introduced by these prepositions as augmented absolutes. hese are not to be confused with free adjuncts where the prepositions imply some form of possession associated with the matrix subject (Stump 1985, 294-5):1
(3)	a. With the discussion completed, the chairman adjourned the meeting for half an hour. b. With a girl in every port, Harry feels pretty contented. (McCawley 1983, 227) = Having a girl in every port, Harry feels pretty contented.
Finally, though personal pronouns seldom realize subjects in absolutes, such instances reveal another puzzling aspect of these constructions. In bare absolutes pronominal subjects sometimes occur in the nominative and sometimes in the accusative form. Stump's (1985) view on this variation is one-sided, as he suggests the latter form is unacceptable in standard English. In contrast, Jespersen (1965, 48) accepts both options, but suggets that the nominative has been supplanted due to a general dislocation of the feeling for cases. Literary examples like (4a) attest to the fact that the accusative, though more frequent in colloquial speech, has become widely acceptable. Jespersen (1965) supports his view that both forms are active by formalizing the surviving nominative subjects like those in (4b) as resulting from some form of coreference with the matrix subject (46):
(4)	a. I, with whom that Impulse was the most intractable, the most capricious, the most maddening of masters (him before me always excepted) ... (C. Bronte, Villette) b. At the church door, however, they separated, he driving back to the Temple, and she to
_her own house. (A. C. Doyle, he Adventures of Sherlock Holmes)_
1 he semantic value of the prepositional phrase does not fully overlap with that of the participial clause. As Stump (1985, 296) shows, the prepositional phrase is a stage-level predicate (denoting a transient state) and may be associated with a modal reading, whereas the participle - an individual-level predicate (denoting a lasting state) allows no such reading (ibid.): With three-inch claws on each paw, a Bengal tiger would be a match for any lion. = If it had three-inch claws on each paw, a Bengal tiger would be a match for any lion. ^ Having three-inch claws on each paw, a Bengal tiger would be a match for any lion.
he existence of augmented absolutes and the survival of nominative pronominal subjects in bare absolutes, as explained by Jespersen (1965), suggest that absolute constructions are not entirely disconnected from the matrix clause. hough the two connections affect the case form of pronominal subjects, they are regarded as separate phenomena. In search of a more systematic account of the structure of absolutes and the case of their subjects, I examine three attempts at explaining such constructions formulated within the Chomskian generative grammar, a structuralist approach based on a universalist view of language. Rejecting the previous proposals for their non-generalizable assumptions and their inability to account for the nominative-accusative subject alternation, I then formulate my own account of absolute constructions. In keeping with the reductionist goals of the Minimalist Program (Chomsky 2000, 2004), the current framework within the Chomskian approach, I propose the same core structure for all absolute constructions, and argue against the need for an external structural licenser.
he paper is organized as follows. Section 2 outlines the key notions of the Minimalist Program. Section 3 presents three previous Chomskian analyses of absolute constructions with critical remarks on their stipulations and disregard for case dualism with pronominal subjects of bare absolutes. In section 4 I pursue a composite approach to absolute constructions, presenting the case dualism as evidence that absolutes have two possible structures and discussing the implications for Case theory. Section 5 concludes the paper.
2. An overview of the Minimalist syntax
One of the key assumptions of the Chomskian generative grammar is that language is a human faculty which is governed by a set of universal principles. hough valid in any given idiolect, such principles are general and thus allow parametric variation, observable in morpho-syntactic differences between specific languages. Accordingly, the Chomskian approach has been dubbed the Principles and Parameters theory (Chomsky 1981). Another crucial assumption of this framework is that linguistic expressions have two interface levels of representation, where the expressions are interpreted phonologically and semantically, while syntax, the system which provides a basic structure for the two representations, is abstract. hese assumptions have been incorporated into the most recent Chomskian approach, the Minimalist Program,2 which defines language as the optimal solution to interpretation conditions at the two interfaces. As such, its goal is to determine the smallest quantity of mechanisms needed to account for the grammaticality of linguistic expressions, reducing the complexity of some systems suggested within the earlier framework or dismantling them altogether.
One of the systems affected by the evolution of the Chomskian approach is syntax. Rather than a system which forms an initial and a post-transformational syntactic structure, it is described within the Minimalist Program as a system of continuous derivation. In the process of this derivation, smaller items are combined to form larger objects. he simplest units to enter this process are lexical items, which combine to form more complex syntactic objects such as phrases. Chomsky (2000, 102) distinguishes between substantive and functional lexical items. Substantive items belong to four categories, N(ominal), A(djectival), V(erbal), and P(repositional), and are the carriers of the core semantic content. Conversely, functional items lack independent semantic content, and expand on the value of substantive elements. he core functional categories are C(omplementizer), T(ense), light v(erb), members of the extended verbal category, and D(eterminer), an extension of the nominal category. Apart from these, other functional categories have been proposed, such as an aspectual category as the bearer of the perfective semantics in (1). In terms of syntax, extension via
2 he following outline is based on the revised Minimalist Program of Chomsky (2000, 2004) because of its economical advantage over the initial model (Chomsky 1995) in terms of the complexity of syntax.
functional categories amounts to complementation - a functional head D selects as its complement a nominal phrase (NP). Although the number of core verbal categories is greater, their sequence of complementation remains constant - C selects a TP, T selects a vP,3 and v selects a VP. he selection of complements by functional elements is referred to as functional selection (f-selection) by Abney (1987, 38). In contrast, the complementation of substantive elements, where the head and its complement semantically describe distinct objects, is called categorial selection (c-selection). his type also depends on the properties of the head, so that a preposition always selects a nominal object, whereas a verb may select either a nominal, a prepositional, or a clausal complement.
Categorial and functional properties can thus be said to be a driving force behind the computation. In abstract syntax, such linguistic properties are represented by features associated with lexical items. Chomsky (2000, 134) suggests that the selectional feature must simply be "satisfied" to enable the operation Merge. his operation combines pairs of elements to form new syntactic objects, whose categorial label and features are identical to those of the respective head (the new object is a projection of the head). hough Adjunction also combines pairs of elements in this way, it is optional, because it does not involve selection. he binary nature of Merge and Adjunction can be seen in the structural representation of the that-clause below:
(5 ) Jones based her judgement on the fact that yesterday Indonesia crushed Malaysia ... (hTJn^ In.de])en.d.en.t, electronic edition)
... CP			
C TP			
that yesterday TP			
Indonesia.	T'		
T	vP		
	t.	v'	
	v+V		VP
	crushed. J		
		V	Malaysia
		t. J	
he schema above is simplified in that it includes only labels of verbal categories. he structure it represents is derived through consecutive applications of Merge and Adjoin, starting with the merger of the verb crushed and the nominal phrase Malaysia. During the course of the derivation, items like the lexical verb crushed and the subject Indonesia are copied and their copies adjoined and merged to their attractors, respectively. Because only the copies are preserved when the structure is transferred to the phonological interface, these mechanisms are regarded as movement.4 In the case of the subject, movement is triggered by the EPP feature on T which is uninterpretable at the interface and must be therefore deleted to prevent the ungrammaticality of the structure. Its deletion is enabled by the operation Agree, whereby the EPP feature probes the contents of the
3	In the earliest stages of the Minimalist Program, the light verb is attributed only to structures with transitive verbs. However, Chomsky (2004) has come to regard all verbal structures which contain a TP as also containing a projection of the light verb, marking the latter as v* where it selects its own argument (typically an agent). To simplify syntactic representations, all instances of light verbs in the present article are marked only as v.
4	he movement of verbal heads will not be explored further, so readers are referred to Chomsky (1995, 133-8) for more details.
structure it has merged with for the nearest matching interpretable feature. Locating its counterpart on the subject, the EPP feature attracts the subject and is then checked and marked for deletion.
Another uninterpretable feature is abstract case (henceforth Case), a remnant of the Principles and Parameters theory, where it was the driving force of movement, since it had to be assigned to nominal projections to license their presence in the clausal structure. hough related to morphological case, it is not necessarily realized by a case inflection. For example, noun subjects in simple English clauses lack a nominative inflection, but are construed as having the nominative Case by analogy with pronominal subjects. Within Minimalist syntax, both subjects are merged into the clausal structure with their Case unchecked and unvalued. Because Case has no matching counterpart, it is checked and provided with a nominative value as a reflex of agreement between the ^-features (gender, number, and person features), uninterpretable on T, and interpretable on the subject. he Case of the object is valued similarly, its accusative value being due to agreement between the ^-features of the light verb v and those of the object (Chomsky 2000, 123). he checking of both Cases is exemplified below with the nominative Case feature listed under the original position of the subject, since checking occurs at the earliest opportunity:
(6)	[cp 0 [tp Ii [ 0 [vp	[askedj [yp her tj for some personal help]]]]]]. u^ ^ i^, uCase[nom\ u^ ^ i^, uCase[ö:cc]
he valuation of Case features shown above can only take place when the uninterpretable ^-features (u^) on verbal heads are complete. As Chomsky (2000: 102) suggests, v has a full set of ^-features when the lexical verb is transitive and active, while T has a full set of ^-features when its projection is selected by C. he relation between the ^-features of C and the Case of the subject can be seen in the fact that infinitival clauses in English typically contain a specified subject when they are introduced by for, as below:5
(7)	It is difficult [Cp *(for) [tP himk [ to [yp t^,	stay calm]]].
u^--------u^ ^ i^, uCase[acc]
he influence of the complementizer for in (7) can furthermore be seen in the obligatory use of the objective form when the subject of the infinitival clause is realized by a personal pronoun. In terms of morphology, the relation between the complementizer and the subject in (7) therefore resembles that between the preposition for and its complement. While the preposition values lexical Case on its complement, realized by semantics-related morphological case in languages with a richer inflectional system, pronominal complements of preposition in English occur in the objective form.
3. Previous generative accounts of absolute constructions
3.1 Pre-Minimalist accounts of absolutes
As mentioned in the Introduction, absolute constructions display a considerable structural diversity, which makes finding a suitable Minimalist description of them challenging. he two accounts presented here were formed within the Principles and Parameters theory, though their proposals are well-suited for Minimalism due to their reliance on basic syntactic devices. In keeping with the overview of the previous section, the outlines provided below are adapted to the Minimalist syntax.
Rather than attempt to provide a comprehensive analysis of absolute constructions, Reuland (1983) focuses on the properties of the -ing constructions. He argues that these are clausal gerunds, as they
5 he placement of the asterisk outside the brackets surroundingfor signifies that omission offOr causes ungrammaticality.
share some characteristics of clauses and some of nominal elements, though they do not fully belong to either category. heir nominal nature is evident where the -ing construction occurs as the complement of a prepositional verb, while the possibility of including non-argument subjects points to their clausal status (Reuland 1983, 108-9):6
(8)	a. You may count on [it raining tomorrow].
b.	*YOu may count on [it to rain tomorrow].
c.	* You may count on [its raining tomorrow].
(8b) is ungrammatical due to the requirement that the complement of a prepositional verb be valued for Case, a feature that the infinitival clause lacks. On the other hand, the nominal gerund in (8c) is also ungrammatical, because its subject is valued for possessive Case, an instance of lexical Case, but receives no semantic role from the head raining. To account for the grammaticality of (8a), Reuland (1983) suggests that clausal gerunds contain a special type of Tense head:7 a nominal -ing which bears a Case feature. he subject in a clausal gerund is then said to receive its Case from the -ing head through percolation or spreading of the Case, in the same way the Case of a Determiner spreads to its complement NP. In augmented absolutes, the Case on -ing is thus said to be valued by the preposition, which Reuland (1983) analyses as a complementizer on a par with for in (7). He comes to this conclusion by assuming that the clausal gerunds in examples like (8a) are CPs as well, treating the preposition on as a complementizer. As proof, he (1983, 132) offers sentences like:
(9)	*John is the one whoi I'm counting [cp [[on] [tjj]] [tp ti2 marryingher\\.
Reuland (1983) argues that the movement of who in (9) is disallowed, because the complementizer on is required to be adjacent to the verb counting (to act as its intermediary in Case valuation), while the subject who must also somehow assume the peripheral position in the CP (represented by ti1). his can only be accomplished if the two are merged, so that they both occupy this position. However, before who can move out of the TP, its Case must be valued by the complementizer, which can only happen if the complementizer merges with the TP in question. Due to these irreconcilable requirements the derivation in (9) crashes. Still, Reuland's (1983) example is not the best choice to show that gerundial clauses are CPs, since it is generally accepted in Minimalist syntax that the trace of a WH-element is merged with C and is accordingly closer to the matrix verb. Another problem of Reuland's (1983) analysis is his view that bare absolute constructions are also CPs, whose null complementizer cannot value the Case of the subject. Based on observations of descriptive grammars regarding the frequent use of the subjective form of pronouns in the role of subjects in bare absolutes, Reuland (1983) resorts to the stipulation that the Tense head of such absolutes values nominative Case. Finally, he argues that pronominal subjects of gerunds frequently occur in the objective form as a side effect of percolation being an indirect means of Case valuation. He (1983, 125) supports this claim with the following examples, where the morphological accusative is possible, since the pronouns are embedded within a conjunction phrase and do not undergo Case valuation themselves.
(10) a. Him and me are going to
b. He and I are going to the party.
Like Reuland (1983), Hantson (1992) also considers absolute constructions to be clauses and analyses them as CPs. However, his proposal is distinct in two ways. First, he (1992, 87) believes accusative
6	For further details, readers are referred to Reuland (1983, 107-9).
7	he Tense element is a category of the Minimalist syntax, so Reuland (1983) discusses its forerunner - the inflectional element (Infl or I). Also, while he makes clear references to inflectional and complementizer heads, his structural analyses label their phrasal projections as S and S', respectively.
Case to be the only choice for the subject of absolutes, claiming pronouns in the subjective form to have no place in productive patterns, only occurring under the influence of prescriptive grammar. Second, his analysis encompasses all absolutes, and he argues that they all contain a verbal head. his head is overt where the Tense8 is associated with non-finite morphology (i.e. with the infinitival to). Where the Tense head is realized by a null element (in the case of -ing,,9 -ed, and non-verbal absolutes), this triggers a deletion of the verb be. Under Hantson's (1992) analysis, the absolute constructions in (11a) and (11b) would therefore have virtually identical structures prior to be-deletion.
(11)	a. [cp 0 [tp he discussion [[t [0] [be-]] [yp tj completed]]]] b. [cP 0 [tp Him [[t [0ing] [beingi]] [yp ti a lawyer]]]]
he null Tin (11a) diflers from that of (11b), as the latter contains a feature associated with gerundial morphology,10 which preserves the auxiliary after it has adjoined to T. In contrast, the null T in (10b) lacks the appropriate feature, resulting in the deletion of the copula after its adjunction to T. Although the comprehensive nature of this analysis has its merits, it also creates a number of difficulties. One of these is represented by infinitival absolutes, where T cannot check the accusative of the subject unless the TP is selected by an overt complementizer. Hantson (1992) dismisses the significance of this by viewing infinitives as diminishing in use and acceptance.11 Regardless of this, his assumption that the T of absolutes checks accusative Case contrasts with the established view that a non-finite T cannot do so by itself. he second problem is the issue of be-deletion, which is tied to the stipulated null T of absolutes, because it again contrasts with the general Minimalist notion that elements are deleted only where their identity can be recovered from the context, as in:12
(12)	James Baur is part of a musical family: his father is a composer, his mother (is) a flutist.
he deletion of the verb in the second clause is an example of gapping, which is made possible by the parallel structure of the two copular clauses. With absolute constructions there is no such parallel, so the presence of be in the structure of absolutes is stipulated, rather than inferred. he final problem of Hantson's (1992) system has to deal with the choice between bare and augmented absolutes. In the majority of cases, the null complementizer and with are interchangeable. As Riehemann and Bender (1999, 477) show, however, the overt complementizer is generally required where the absolute construction involves a predicative idiom with a DP predicate:13
(13)	*(With) peace talks old hat, it's hard to get a sense of hopefulness in the Middle East these days.
8	Like Reuland (1983), Hantson (1992) also uses the categorial labels of the principles and parameters framework (see note 9).
9	Also like Reuland (1983), Hantson (1992, 80) assumes the -ing affix to be associated with tense in the same way the infinitival to can realize a Tense head. he latter is only possible, because to is a lexical item, however. he affix -ingis not, and is rather meant to represent a kind of gerundial feature, a combination clausal (Tense) and nominal (Case) properties.
10	Hantson (1992) and Reuland (1983) assume that the head of TP (or S) is not null, but actually contains the gerundial morphology, which triggers the movement of the verb to T (I/Inß). his view is incompatible with the Minimalist syntax, where verbs enter a derivation fully inflected, and are then attracted by functional heads. Only when combined with specific heads, can their value be interpreted (e.g. the transitivity of a verb is only interpreted after it has adjoined to v).
11	Hantson (1992, 91) does this by showing that infinitival absolutes are deemed progressively less acceptable when containing passive predication, expletive subjects or when introduced by without:
(i)	With Jane to look after his children, his future was looking brighter.
(ii)	?With his children to be looked after by Jane, his future was looking brighter.
(iii)	With there now to be someone to look after his children, his future is looking brighter.
(iv)	*Without there soon to be someone to look after his children, his future would be very gloomy.
(v)	* After listening to the weather forecast, we decided that, with it to rain all day, we had better stay at home.
12	For more details on gapping and ellipsis, readers are referred to Smith (2001) and the references listed there.
13	Riehemann and Bender's (1999) conclusion is based on a brief survey of acceptability of with and with-less idiomatic absolutes. Since their findings are formalized within the Head-Driven phrase Structure Grammar, they are not examined in further detail.
3.2 A Minimalist account of absolutes
he fact that absolute constructions involve a form of predication motivated Reuland (1983) and Hantson (1992) to analyse them as clauses, though both had to resort to stipulations regarding some of the Case properties of absolutes. For Hantson (1992), non-verbal absolutes also present a problem, because he views them as clauses which have undergone ^^-deletion in a manner that has no parallels in Minimalist syntax. Rather than assume that the auxiliary enters derivation and is then deleted, Lundin (2003) argues in favour of a structural analysis that does not need to include the auxiliary in the first place. Such a construction is referred to as a small clause.
In her analysis of Swedish absolutes, Lundin (2003) expands the analysis of small clauses suggested by Stowell (1981), who argues that they are not a uniform category, but projections of non-verbal heads. Although Lundin (2003) agrees that the substantive category of a small clause matches that of its predicate, she redefines small clauses to include participial predicates. Assuming that these project a vP1415 (the simplest structure to involve a subject position), she proposes that the other small clauses also form light functional projections of substantive categories, as below:
(14)	a. 73 people have been drowned in the area, [nP [most of them] [np children]].
b.	73people have been drowned in the area, [ap [most of them] [ap still young]].
c.	73 people have been drowned in the area, [pp [most of them] [pp still in their teens]].
d.	[vp Her aunt having left the room], I asked Ann for some personal help.
By suggesting that vP and nP have parallel structures, however, Lundin (2003) regards the predicative element of the latter as an NP. his is problematic, since it means that any articles within this element would be located within an NP, whereas they should occur as the head of a DP. Lundin (2003) also adopts Stowell's (1981) views regarding the Case of the subject in small clauses, which is said to be valued by an external head. his is obviously not possible if the small clause analysis is directly applied to bare absolutes, so Lundin (2003) assumes that all absolutes are prepositional phrases containing either an overt or a null preposition, which values the Case of the subject in the small clause. According to Quirk et al. (1985, 1124), the prepositions which introduce small clauses assign their complements the semantic value of an accompanying circumstance. Lundin (2003, 149) reinterprets this by suggesting that the prepositional head bears an interpretable Tense feature, which is comparable to the one found on the infinitival Tense head to.16 hese features anchor the predication to the temporal frame of the matrix T.17 hus, the "accompanying" value of the small clause roughly corresponds to simultaneity with the matrix proposition (when the small clause denotes a temporary state) or to a perpetual circumstance (when the small clause denotes a property). he two interpretations are showcased by (15a) and (15b), respectively.
(15)	a. Without her being conscious of it at the time, her struggle for independence rotated around the figure of Hansel. (J. Casterton, Creative writing. A practical guide)
b. And him being a lawyer and all, we cooked up a grand scheme between us. _(E. Richmond, Love of my heart)_
14	In Lundin's (2003) approach, intransitive participles are also understood to project a vl^.
15	It should be noted that Lundin's (2003) analysis of small clauses is not the sole option within the Minimalist framework. hus, Al-Horais (2007) suggests that the core structure of bare absolutes is a TP, which may contain any of the predicative phrases, which is problematic, because T otherwise only ever selects a vP. Al-Horais (2007) also lists other generative accounts of small clauses, though an overview of these is beyond the scope of the present paper, so readers are referred to his article for further details.
16	he fact that this element is of prepositional origin supports this analysis.
17	he temporal interpretation of infinitival clauses also depends on the properties of its selector:
(i)	Ian has decided to pursue other activities. (the infinitive has the irrealis and posteriority reading)
(ii)	I'm glad to see you: Yet Im sorry too. (the infinitive has the realis and simultaneity reading)
Considering the examples of absolute constructions listed in the previous sections, however, Lundin's (2003) analysis has two weaknesses. First, the assumption that the preposition with may select a small clause of any category defies the c-selectional properties of prepositions, which typically accept nominal complements. Secondly, prepositions value a non-nominative Case, so the suggestion that all absolutes have the structure of PPs entails the dismissal of subjective form pronouns as subjects in bare absolutes.18
4. A composite account of absolute constructions
So far, I have reviewed proposals claiming that all absolutes have the same syntactic structure (notwithstanding the distinctions of inner structure in Lundin's (2003) account). A common feature of these accounts is the suggestion that bare absolutes are headed by null elements, which has been shown to be problematic for depending on stipulations regarding the nominative form. hus, Reuland (1983) merely observes that the nominative subject occurs when the absolute contains a null complementizer, while Hantson's (1992) and Lundin's (2003) analyses reject nominative subjects altogether. However, since Reuland (1983) focuses solely on clausal gerunds, and Hantson (1992), though briefly discussing even infinitival absolutes, relies on an ad hoc rule of be-deletion to explain verbless absolutes, Lundin's (2003) account seems the most promising for its comprehensive economy. he main issues of her approach result from the nature of the prepositional head, which should disallow non-nominal complements and the nominative subject within the small clause. To circumvent these problems, I propose a modification of Lundin's (2003) account whereby small clauses are projections of a category-neutral head and may occur in unselected positions.
4.1 he categorial status of absolutes
Lundin's (2003) suggestion that the core predication of absolutes is a small clause can be salvaged if small clauses are analysed as projections of a head without inherent categorial features. An example of such an element is the conjunction and, whose lack of categorial features allows it to assume the characteristics of objects with which it is merged. he conjunction phrase may therefore only be an object of a preposition when it contains nominal, but not verbal (clausal) or prepositional elements:
(16) a. Europe unites in austerity protests against [cuts and job losses]. (he Guardian, electronic edition)
b.	*Europe unites in austerity protests against [that salaries are being cut and that people are losing their jobs].
c.	* Europe unites in austerity protests against [with lower salaries and out of a job].
Unlike the conjunction and, however, the head of a small clause must be able to select constituents of distinct categories. he item in question is called the depictive, a non-core functional item first suggested by Pylkkänen (2002) as the head in depictive secondary predication. In her terms, the depictive phrase (DepP) consists of a head and a predicative phrase, and is merged into the clausal structure just prior to the DP with which it is associated. A subject depictive thus combines with v', and the resulting second v' is merged with the subject (before the latter is copied and merged in TP), as below (Pylkkänen 2002, 29):'9_
18	To my knowledge, the only instances where the complement of a preposition is realized by a personal pronoun in the subjective form are marked due to the influence of the relative clause, so these cannot be invoked to support Lundin's (2003) proposal: Temporal and causal sequences are ordered retrospectively by us, i.e., by we who have some idea of the development we seek to organize. (D. A. Duquette, Hegel's History of Philosophy: New Interpretations)
19	According to Pylkkänen (2002), the DepP in (17) actually combines with a projection of voice, which is comparable to light v as the bearer of the transitive feature which checks accusative Case.
(17)	Sue saw Peter tired, [^p Suej [^' T [vp tj [v' [v' saw Peter] [pepP tired DEP] ]]]]
he original proposal is amended by Marušič, Marvin and Žaucer (2008), who show that the depictive and its associate do not enter the derivation as separate objects, but are merged so that the DP is a constituent of DepP. Instances of secondary predication in prepositional phrases, though rare in English, seem to support this conclusion and are said to have the structure represented below:
(18)	he bra^n surgreon iiaidUto operate on the: p^t:ient:i wid^-a^t^akiei. (Maling 2001, 424)
PP
on	DepP
the patient Dep'
Dep wide-awake
According to this analysis, the DP functions as the "host", because it is targeted for Case valuation and the assignment of the semantic role by external items, while the depictive itself receives neither of these. It only assumes its host's nominal feature, which enables it to merge in a Case position. hese traits are not unique to free adjuncts, however. hey are shared by small clauses with a concurrent reading, the only exception being the assignment of the semantic role by an external selector (which targets the entire predication, rather than the subject entity alone). he idea that small clauses of this kind can be analysed as DepPs is supported by a number of syntactic parallels. Like the free adjunct depictives, small clauses can occupy a position which is not associated with Case. A typical example of this is passivization, where the host and the subject are forced to move out of the DepP and into a position that is accessible to nominative Case valuation.
(19),.,	salsify can be eaten raw in salads or cooked, (IdealHome, electronic edition) (cf. People often eat salsify raw.)
(20)	John is considered very stupid / a nice guy.
Although the DepPs in the previous examples contain a null depictive head, its position can also be occupied by an overt item. his item can be found in both free adjunct depictives and small clauses when they have a qualifying meaning and a nominal predicate:20
(21)Not	until much later was a similar vegetable ... served [DepP as a side dish] ... (he National Trust Magazine, electronic editian)
(22)I'd	long considered [pepP him ... as a man with a severe personality disorder] ... (he Telegraph, electronic edition)
he majority of free adjunct depictives contain either an adjectival or a nominal predicate, but they can also contain prepositional and verbal predicates. Such a categorial range of predicates can also be observed with small clauses, where the verbal predicate is predominantly participial in nature.21 However, while the infinitive can appear as the predicate in free adjunct depictives, it
20	he depictive head in small clauses can also be overt with non-nominal predicates:
Eliot... is able to see his childhood as containing ... he Elementary Forms of the R^eligious Life. (R. Crawford, he savage and the city in the work of T S Eliot)
21	Like Lundin (2003), I assume that the -ingpredicate of small clauses is participial and that v is the locus of participial morphology. By assuming that the depictive head borrows the nominal nature of its host or subject, I dispense with the need for a gerundial -ing along the lines of Reuland (1983) (for an alternative view of gerundial structure, which could represent a counterpart to my
is rarely interpreted as implying an arrangement for the future like the one used in (2c). Instead, the infinitival predicate in post-verbal positions is mostly interpreted as either an adverbial clause of purpose or as an object clause with a simultaneous interpretation typical of non-verbal small clauses. he preference for a non-posterior interpretation of the infinitival predicate in such positions is showcased by parallels between examples in (23) and (24):
(23)	a. Sue saw Peter tired / as a concerned friend / in high spirits / wearing her new dress. b. I consider John very stupid / a nice guy / out of order / as lying too much.
(24)	a. Sue saw Peter (anyway) to leave.
= Sue saw Peter (anyway), so she could leave. ^ Sue saw Peter (anyway), as she was set to leave. b. I consider John to be very stupid.
= It is my opinion that John is "v'ery stupid.
^ It is my opini(3i^ thiat Jolsn wheill lue very stupid.
As such, the non-argumient ^bso^uite ct^nsti-uction appears to be the only clear instance of a small clause with an infinitival predicate wh^hich imjjlies a posterior arrangement. Such an interpretation is also unambiguous in a small number of free adjuncts, mainly those which occur in the clause-initial position and contain about:
(25)About	to be leave, Sue saw Peter (anyway).
he scarcity of infinitival predicates is therefore not detrimental to the depictive analysis of absolutes outlined in this section. Rather, the unavoidable issue with suggesting free adjuncts and small clauses share the same structure has to do with the transparency of depictives for external operations. As discussed above, only free adjunct depictives allow their subjects to be assigned a second semantic role by an external head, otherwise subjects of both free adjuncts and small clauses are accessible for Case valuation and movement, which is presented in (19) and (20). his distinction can be attributed to a semantic difference between the depictive head used in free adjuncts and small clauses, respectively, with the only the latter requiring a semantic role of its own. he assignment of a semantic role to the depictive therefore takes precedence over the assignment of another semantic role to the subject of the small clause. Finally, the crucial similarity between the two types of depictives is seen in the fact that both borrow the categorial status of their subject, which enables them to occur as complements of prepositions. An example of such a small clause can be seen within the prepositional free adjunct in (3b), while an example of a free adjunct inside a PP can be seen in (18). Having thus established the advantages of the DepP analysis of small clauses, I propose that bare absolutes are DepPs, while augmented absolutes are PPs with DepP complements. hese are showcased by the representations of absolutes in (26a) and (26b) below:
(26)	a. he discussion completed ...	b. With the discussion completed ...
DepP	PP
The discussion Dep'	With DepP
Dep completed	the discussion Dep'
	Dep completed
DepP-analysis, see Abney (1987, 151ff)).
he structural distinction between the two groups is corroborated by the fact that subjects realized by personal pronouns obligatorily occur in the objective form only in augmented absolutes. Conversely, the dualism of subject forms in bare absolutes suggests that the Case value of the subject is neither due to a syntactic relation, like the one between the complementizer for and the subject in infinitival clauses, nor to lexical properties of a prepositional selector. However, such a conclusion is at odds with the stipulation that all nominal projections bear an uninterpretable Case feature, which I argue in the following subsection to be superfluous.
4.2 Absolute case versus abstract Case
Any discussion of Case in absolute constructions is based on observations regarding the case form of their subjects. In this, I assume Jespersen's view that both the subjective and the objective forms are available with pronominal subjects in bare absolutes, though the latter one is prevalent. According to Jespersen's (1965) proposal that the subjective form is mainly manifested when the subject of the absolute is coreferential with the matrix subject bare absolutes must be transparent to external influence. Nevertheless, the following example serves as evidence that coreference is not needed to trigger the subjective form of the pronominal subject:
(27)	Once, she being childishly curious, her father had given her a sip of the tot of rum he had
bought to ease her grandfather's chest. (E. Rhodes, Ruth Appleby)
Such examples further show that the case of the subject in bare absolutes cannot be universally determined through syntactic relations, which is not necessarily a drawback. For one, morphological case does not always reflect abstract Case. Also, due to the adverbial status of absolutes, the case of the subject could be attributed to the semantic and morphological factors rather than syntactic or lexical ones. A similar instance of semantic case morphology could be attributed to nominal adjuncts, such as these days in (13), which also appear to lack a structural licenser. he fact that nominal adjuncts do not cause the derivation to crash despite the lack of ostensible licensers suggests such nominal projections do not have a Case feature in need of valuation. A similar conclusion is reached by Boskovic (2006), who shows that a preposition which values instrumental Case on its complement in Serbo-Croatian is not a suitable choice, if the adjunct is to convey the meaning of direction (op. cit.: 530):
(28)	a. Trčao je šumom. (He ran through a/the forest.)
b. Trčao je *s(a)/kroz pet šuma. (He ran through five forests.)
he adjunct sumom in (28a) has instrumental morphology, and signifies direction, while the adjunct in (28b) consists of a nominal in the quantificational genitive form and thus requires a preposition to convey the same meaning. he preposition which values instrumental Case s(a), however, is not semantically suitable, and it is kroz, which values accusative Case, that enables the interpretation listed in the brackets. Since šumom in (28a) is not in accusative form, Boskovic (2006) suggests that such nominal adjuncts are bare and that their Case feature is semantically interpretable. his conclusion is a further departure from the Case theory of the Minimalist Program, which identifies Case as the only feature without an interpretable counterpart, whose valuation is a reflex of agreement between ^-features.
Schütze (2001)22 also opposes the idea that all nominal projections bear uninterpretable Case, arguing that even argument DPs can enter the derivation Caseless, and have their form determined
22 Although Schütze (2001) admits his analysis is better suited to the theory of Distributed Morphology, it can also be seen as complementary to the purely syntactic approach of the Minimalist Program.
through other devices. One of these is case matching,23 which makes a Caseless DP mirror the case form of one which has been checked for Case, if these elements are coreferential. he other is what Schütze (2001, 221) calls default case, a form that is either "least featurally specified" or simply used to realize Caseless DPs. In English, the default case is accusative and can be seen on DPs who are in - Reuland's (1983) terms - not adjacent to its Case-checking selector (as with conjunction phrases, left dislocation, and gapping), and on predicative pronouns:
(29)	a. Me / *I, I ^ike bea^ns.
b. We can't eat caviar and him / *he (eat) beans.
(30)	... I hope I've paid him due credit, thought he did start that nasty rumour that he was me or
I was him. (She Magazine, electronic edition)
As shown by (29), case matching is weaker than default case in English. Furthermore, due to the scarce case system of English, the common case of DPs which are used as temporal adjuncts can also be said to be an instance of default case.24 his accounts for the objective pronominal subjects in bare absolutes, though it still leaves the issue of subjective pronominal subjects unresolved. According to Jespersen (1965), these are a remnant of an earlier dominant pattern, when the subjective case was the default form in English. heir survival may be due, in part, to coreference with matrix subjects, bearers of nominative Case, which suggests that subjective forms in such instances are determined through case matching. heir use is not restricted to such instances, however, and it is irrelevant whether this is a reflex of prescriptive grammar, as argued by Hantson (1992), since the pattern is still productive, albeit no longer prevalent.
Acknowledgement of subjective pronominal subjects in bare absolutes is comparable to arguments in favour of semantically determined and default case forms in that these phenomena point to a gap in the Case theory. Nevertheless, attributing interpretable Case to the subjects of bare absolutes or suggesting them to be exceptional for not having Case does not resolve the issue of dissociation between abstract Case and morphological case. Since neither of the two proposals offers a definitive generalization regarding the relationship between two, I follow SigurSsson (2009) in taking the aforementioned dissociation as evidence that abstract Case, an exceptional feature to begin with, does not exist. Instead, case is a purely morphological device, whose value depends on the interplay of lexical, syntactic, semantic factors with the morphology of individual languages. his conclusion is desirable from a minimalist perspective, since it removes several exceptions, which could not be explained without further, non-generalizable stipulations. hese include the indirect method of Case valuation and the non-relational determination of case with nominal adjuncts and the bearers of default case.
Finally, the removal of Case from syntax is desirable, since its original roles, triggering movement of nominal projections and licensing their presence in the structure, have been assumed by other mechanisms in the Minimalist Program. As shown in section 2, movement of nominal projections is triggered by the EPP feature. Licensing of nominal projections, however, is already regulated by the Full Interpretation principle, which requires that only elements which contribute to the interpretation be present in the structure. As such, it prohibits the adjunction or merging of nominal projections which cannot be assigned a semantic role at the interface. Since semantic factors and case morphology have been shown to be connected in the case of nominal adjuncts, this also explains why in most cases English absolutes with nominal predicates must be selected by an overt preposition. his requirement is due to the poor morphological system of English, which is unable to distinguish between nominal
23	his mechanism should not be confused with feature matching under Agree.
24	Conversely, in Latin and classical Greek subjects of absolute constructions typically occur in case forms associated with adverbial functions (ablative and genitive, respectively), shared by any case-bearing predicative elements.
adjuncts on the basis of their meaning, as opposed to languages like Serbo-Croatian. As such, nominal adjuncts are licensed either for their semantics, specifically when they refer to time.25 On the other hand, bare absolutes with nominal predicates can only be licensed through their distributive coreference with the matrix subject, as in (4b). Beyond this, English morphology does not facilitate an adverbial interpretation of the absolute construction or the relation between its subject and the nominal predicate. In compensation for this deficit, English absolutes with nominal predicates are obligatorily augmented, so that the preposition enables the predication of such absolutes to be interpreted as an accompanying circumstance of the matrix predication. Surprisingly, examples like (31) seem to be exempt from this requirement:
(31) Anita wears a Tibetan scarf and high grey leg-warmers, her hair a tumble of untamed curls. (A. L. Hall, Deliria)
Although the small clause in (31) has an adverbial function, it is closely associated with the matrix subject. his connection leads me to believe such instances actually contain free adjuncts whose predicate is headed by a non-overt preposition or a participle, comparable to those in (3b).
5. Conclusion
In this paper, I have examined several options of accounting for the case of the subject in English absolute clauses and its correlation with the category of these constructions within the Minimalist Program. I have shown that, contrary to descriptive grammars, Minimalist accounts tend to favour the accusative absolute, striving for a common structural analysis of absolutes, which becomes the source of difficulties for each proposal. Hantson (1992) and, in a limited way, Reuland (1983) consider absolutes to have the structure of clauses, which is problematic for stipulating unique occurrences of mechanisms like Case checking by a non-finite Tense and be-deletion. Lundin (2003) advocates a small-clause analysis with four categorial subtypes, undermined by her assumption that all absolute small clauses are selected by a preposition. his is irreconcilable with categorial distinctions between small clauses due to selectional restrictions of prepositions, and due to the resulting dismissal of any instance of a nominative subject within absolutes. Finally, I have proposed a mono-categorial small-clause analysis of absolutes as depictive phrases, whose case is determined through an interaction of syntactic and non-syntactic factors. Despite the semantic differences between depictive secondary predication and small clauses, I have shown the analysis to be viable since the two behave alike in many circumstances. Furthermore, the proposal that case dualism in bare absolutes is enabled by the unselected status of such constructions and the corresponding rejection of abstract Case are shown to be advantageous. Not only does the dismissal of Case remove an exception from syntax, but its role as a licenser of nominal elements can readily be shifted to semantic and morphological mechanisms in the case of nominal adjuncts. It remains to be seen whether such mechanisms suffice as licensers of nominal elements in other functions.
References
Abney, Steven P. 1987. "he English Noun Phrase in its Sentential Aspect." PhD diss., MIT.
Al-Horais, Nasser. 2007. "he categorical status of the small clause node: A Minimalist approach." Newcastle Working Papers in Linguistics, Volumes 12 & 13: 96-108. Newcastle upon Tyne: University of Newcastle.
Boškovic, Željko. 2006. "Case Checking versus Case Assignment and the Case of Adverbial NPs." Linguistic
Inquiry 37 (3): 522-33._
25 McCawley (1988) shows that the choice of premodifiers influences the ability of temporal nominals to occur as nominal adjuncts, though he regards them as PPs with a null head (588): (i) We went inhere t^he same day. (ii) We went the-re *(on) a subsequent day.
Chomsky Noam. 1981. Lectures on Government and Binding, he Pisa Lectures. 7th ed. (1993). Dodrecht: Foris.
---. 1995. he Minimalist Program. Cambridge, Massachusetts: he MIT Press.
---. 2000. "Minimalist Inquiries: he Framework." In Step by Step: Essays on Minimalist Syntax in Honor of
Howard Lasnik, edited by Roger Martin, David Michaels, Juan Uriagereka, and Samuel Jay Keyser, 89-155. Cambridge, Massachusetts: he MIT Press.
---. 2004. "Beyond Explanatory Adequacy." In Structures and Beyond, he Cartography of Syntactic
Structures, Volume 3, edited by Adriana Belletti, 104-31. New York: Oxford University Press.
Hantson, Andre. 1992. "Case Assignment and BE-deletion in Non-finite Clauses, with Special Reference to (Absolute) Free Adjuncts." Belgian Journal of Linguistics 7: 75-94. Namur: Facultes Universitaires Notre-Dame de la Paix.
Jespersen, Otto. 1965. A modern English grammar: on historical principles. Part V, Syntax. 4th volume. London: George, Allen & Unwin.
Lundin, Katarina. 2003. "Small Clauses in Swedish: Towards a Unified Account." PhD diss., Lund University.
Maling, Joan. 2001. "Dative: the heterogeneity of mapping among morphological case, grammatical functions, and thematic roles." Lingua 111 (4-7): 419-64. Amsterdam: Elsevier B.V.
Marušič, Franc, Tatjana Marvin, and Rok Žaucer. 2008. "Depictive secondary predication with no PRO." In Formal Description of Slavic Languages: he Fifth Conference, Leipzig 2003 (Linguistik International), edited by Gerhild Zybatow, Luka Szucsich, Uwe Junghanns, and Roland Meyer, 373-92. Frankfurt am Main: Peter Lang.
McCawley, James D. 1983. "What's with with?" Language 59 (2): 271-87. Baltimore: Waverly Press.
---. 1988. "Adverbial NPs: Bare or Clad in See-hrough Garb?" La^nguage 64 (3): 583-90. Baltimore:
Waverly Press.
Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech, and Jan Svartvik. 1985. A Comprehensive Grammar ofthe English Language. London: Longman.
Pylkkänen, Liina. 2002. "Introducing Arguments." PhD diss., MIT.
Reuland, Eric J. 1983. "Governing -ing!' Linguistic Inquiry 14 (1): 101-36. Cambridge, Massachusetts: MIT Press.
Riehemann, Susanne Z., and Emily Bender. 1999. "Absolute Constructions: On the Distribution of Predicative Idioms." In Proceedings ofthe 18th West Coast Conference on Formal Linguistics, edited by Sonya Bird, Andrew Carnie, Jason D. Haugen, and Peter Norquest, 476-89. Sommerville, MA: Cascadilla Press.
Schütze, Carson T. 2001. "On the Nature of the Default Case." Synl^ax 4 (3): 205-38. Oxford and Malden: Blackwell Publishers.
SigurSsson, Halldor Ä. 2009. "he No Case Generalization." In Advances in Comparative Germanic Syntax, edited by Artemis Alexiadou, Jorge Hankamer, homas McFadden, Justin Nuger, and Florian Schafer, 249-79. Amsterdam: John Benjamins.
Smith, Noah A. 2001. "Ellipsis Happens, and Deletion Is How." In University of Maryland Working Papers in Linguistics 11, edited by Edited by Andrea Gualmini, Soo-Min Hong, and Mitsue Motomura, 176-91. Maryland: Linguistics Department of the University of Maryland.
Stowell, Tim A. 1981. "Origins of Phrase Structure." PhD diss., MIT.
Stump, Gregory T. 1985. he semantic variability of absolute constructions. Dordrecht: D. Reidel Publishing Company.
Visser, Frederik h. 1972. An historical syntax ofthe English language. Part 2, Syntactical units with one verb: continued. Leiden: Brill.