Case Checking versus Case Assignment and the Case of Adverbial NPs

On the basis of genitive of quantification (GQ) in Slavic, I argue in this squib that the Case-checking theory is empirically superior to the Case assignment theory. I also argue that Slavic GQ provides evidence that adverbial NPs (Larson's (1985) bare NP adverbs) are not subject to the traditional Case Filter, and I suggest that the Case of such NPs is interpretable. During the discussion, I also examine how Case-marked modifying adjectives and noninitial NP conjuncts are Case-licensed.

The squib is organized as follows. In section 1, I summarize the basic properties of Slavic GQ. In section 2, I show that GQ can tease apart Case checking and Case assignment. In section 3, I discuss Case properties of adverbial NPs.

1 Introduction: Genitive of Quantification

Examples (1)-(2) show what happens when a numeral NP occurs in a structural Case context in Russian.¹

(1)

1. a. Ivan kupil    odnu      mašinu.
   Ivan bought one(ACC) car(ACC.SG)

2. b. *Ivan kupil    odnu      mašiny.
   Ivan bought one(ACC) car(GEN.SG)

(2)

1. a. Ivan kupil    pjat' mašin.
   Ivan bought five cars(GEN.PL)

2. b. *Ivan kupil    pjat' mašiny.
   Ivan bought five cars(ACC.PL)

With 'one', both 'one' and the noun receive their Case from the verb. However, with a higher numeral like 'five', the noun receives genitive.² This pattern is traditionally interpreted as indicating that only higher numerals have the ability to assign GQ. As for inherent Case contexts, when a numeral NP occurs as an object of an inherent-Case-assigning V, as in (3)-(4), both the noun and the numeral ('one' as well as 'five') bear the inherent Case in question.³

(3) Ivan vladeet odnoj        fabrikoj.
Ivan owns one(INSTR) factory(INSTR.SG)

(4)

1. a. Ivan vladeet pjat'ju        fabrikami.
   Ivan owns    five(INSTR) factories(INSTR.PL)

2. b. *Ivan vladeet pjat' fabrik.
   Ivan owns five factories(GEN.PL)

The descriptive generalization is that GQ overrides structural Case (2), but not inherent Case (4).

There are many analyses of the basic GQ paradigm in the literature. Most authors assume that the reason why GQ overrides structural accusative in constructions like (2) is that the GQ assigner, the numeral, is closer to the relevant noun than the verb. The strategy pursued with respect to (4) is to enforce the satisfaction of the Case property of the inherent-Case-assigning verb independently of the distance factor. Freidin and Babby (1984) and Babby (1985) do this by appealing to the Principle of Lexical Satisfaction, which requires that lexical properties be satisfied at all levels of representation. Assuming that instrumental Case assignment is an idiosyncratic property of the verb that must be specified in the lexicon, it follows that instrumental will take precedence over any other Case-licensing strategy in (3)-(4) (see, however, Franks 1998, 2002 and Bošković, to appear, for some arguments against this analysis). Franks (1994, 1995) presents an analysis based on Chomsky's (1986) Case theory, in which structural Case is assigned at S-Structure and inherent Case at D-Structure, and he proposes that GQ is a structural Case in Russian. 'Factory' then must bear instrumental in (4) because the verb assigns its inherent instrumental Case at D-Structure before the Case-assigning ability of the Q is activated. On the other hand, adopting the gist of Chomsky's (1986) theory of inherent Case (but not requiring the assumption that D-Structure and S-Structure are levels of representation), I argue elsewhere (Bošković, to appear) that GQ cannot override inherent Case because, as Chomsky argues, inherent Case is associated with θ-role assignment; as a result, if an inherent-Case-marking verb fails to check its inherent Case against its object, it will also fail to θ-mark it.⁴ Under this analysis, (4b) ends up violating the θ-Criterion. θ-considerations are irrelevant in a structural-Case-assigning context since structural Case is not associated with θ-role assignment in Chomsky's (1986) Case system. For ease of exposition, Iwill adopt the analysis developed in Bošković, to appear, in the discussion below...