Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.philol.msu.ru/~otipl/new/fdsl/abstracts/biskup.pdf Дата изменения: Sun Nov 9 20:25:46 2008 Дата индексирования: Wed Jan 14 13:01:39 2009 Кодировка:

Prepositional Projections Introduction: This paper deals with Czech and Russian prepositions. Its main objective is to show that prepositional properties are derivable from the following PP structure: (1) [TpP TP (unval-fs, valT-f) [DirP Dir [LocP Loc [DP (AP) DP(val-fs, unvalT-f) ]]]] Data: As shown in (2) and (4), Ps can merge with different categories: with a preposition (2a), (4), with a noun phrase (2b), an adverbial whP (2c), (2g), a verb (2d), an adjective (2e), and a clause (2f). Prepositions can also assign case to different categories, see (2a), (2b), (2e), and (2g). Czech: (2) a. do-ped-u b. do Prah-y c. do-kdy d. do-jМt to-in.front.of-gen.sg to Prag-gen.sg to-when to-go `forward' `to Prag' `till when' `come' e. do-modr-a f. pes-to-ze [TP ...] g. z-kam-a to-blue-gen.sg over-this-that out-where-gen.sg `(a little bit) blue' `although' `from where' Certain prepositions can assign different cases, as demonstrated in (3). There are also complex prepositions, as shown in (4). And some prepositions can bring about speaker anchoring, consider (5). Russian: (3) a. na /v / o stol-e b. na /v / o stol on in about table-loc on in about table.acc (4) iz-za X.gen out-behind X.gen `from behind' Czech: (5) Pavel do-nesl knihy. Pavel to-carried books `Pavel brought books.' Proposal: Complex PPs such as (2a) and (4) show that PPs can be decomposed into two projections: Dir(ectional)P and Loc(ative)P, where the first morpheme spells out Dir and the second one Loc, consider (1). According to Pesetsky & Torrego (2004), structural case is an unval(ued) T-f(eature) on D that enters an Agree relation with T. I extend their proposal and suggest that prepositional cases are also an unvalT-f on D and that Ps, in addition to val(ued)T-f, bear probing unval-f(eature)s on TP, as shown in (1). Both types of features participate in the case-assigning process (Agree between TP and the complement of Loc). As a consequence of Agree, unvalT-f on the complement of Loc is valued by valT-f on TP (i.e. the complement gets a case, as in (2a), (2b), (2e), (2g)) and -fs on TP are valued by the complement (there are languages with overt P agreement, e.g. Irish, Hungarian or Welsh). There is a correlation between the type of the TP complement and the value of the T-f on TP (the type of case). Depending on its complement ­ DirP does not have to be projected, as in (3a) - valT-f on TP assigns either the locative case (3a) or the directional case (3b). The complement of Loc does not have to be only a bare DP (2b), it can also be a CP (sentential DP), as in (2f), which bears -fs and unvalT-f. Note that that-clauses marked with an oblique case typically occur with a demonstrative in Czech (6):

1


b. za-*(tМm)-[C' co [TP ...]] c. pro-*(to)-[C'ze [TP ...]] (6) a. po-*(tom) [C' co [TP ...]] after-this what/that behind-this-what/that for-this-that `after' `while' `because' The complement of Loc can also be a covert DP (2a), which is modified by an adjective (2e), or an adverbial whP containing a DP (2g); cf. Caha & MedovА (to appear), who argue that Czech manner adverbials like rychl-e `fast' are marked with the locative case and contain a covert noun, or Vangsnes (2008), who shows that Scandinavian manner wh-phrases `how' contain an abstract nominal morpheme. Thus, for the complex PP dopedu, I propose the following syntactic structure, where TP Agrees with DP: (7) [TpP TP [DirP do [LocP ped [DP N-u ]]]] When the complement of Loc is not overt (i.e. it is referentially defective; note that although the case on dopedu is a reflection of Agree between TP and the DP, it cannot vary, it is always genitive of the masculine paradigm hrad `castle'), the interpretation must be provided by some other mechanism. Referentiality of PPs is placed in TP because TP bears -fs (it Agrees with the complement of Loc). Given the T nature of TP, I propose that TP can get the appropriate interpretation in other T category, specifically, in the verbal TP. TVP is the place of speaker anchoring because the speech time is located there and also place of referentiality of the clausal subject because unval-fs on TV Agree with val-fs on the subject. Therefore, in (8) Pavel's motion is localized wrt. the clausal subject (Pavel), and in (5) the goal location of books is determined by speaker's spatiotemporal position. (8) Pavel sel do-ped-u. Pavel went to-in.front.of-gen.sg `Pavel went forward.' Ps (with their argument structure and lexicosemantic features) are merged as Loc and then incorporate into Dir (if projected) and into TP. If PP occurs in the complement position of a verb, then the complex P-head can also incorporate into the verb, as in (2d), and makes the verb perfective (arguments supporting the view that verbal prefixes are incorporated Ps will not be discussed here). Here I argue that valT-f on TP is responsible for boundedness (definiteness effects). In (2d), valT-f on the complex P-head incorporated into the verb values unvalT-f on Asp as bounded (for features of Asp see (9)), which brings about perfectivity. (9) [CP C [TP T(unval-fs, valT-f) [AspP Asp (unval-fs, unvalT-f) [vP v [VP V PP ]]]]] Definiteness effects induced by valT-f on TP are also observable in the case of verbal objects with structural accusative because unvalT-f (case) on the direct object is valued by T-f on Asp, which was valued by valT-f on TP incorporated into the verb. Thus, extraction from an object with the perfective structural accusative is worse than extraction from an object with the imperfective structural accusative, for Russian examples see Romanova (2007). Similarly, in (2c) valT-f on do brings about boundedness (the final point) of the time interval, in contrast to the bare kdy `when', which expresses an unbounded interval. It is also known that there is a relation between prepositional cases and definiteness; PPs are islands for extraction. In such cases, valT-f on TP valuing unvalT-f (case) on the prepositional complement induces islandhood (not shown here), similarly as in the case of direct objects. References: Caha, P. & L. MedovА (to appear), Czech Adverbs as Case-marked Adjectives. In: Proceedings of FDSL-7. UniversitДt Leipzig. Pesetsky, D. & E. Torrego (2004), Tense, Case, and the Nature of Syntactic Categories. In: J. GuИron & J. Lecarme (eds.), The syntax of Time. Cambridge, Mass.: MIT Press, 495-539. Romanova, E. (2007), Constructing perfectivity in Russian. Ph.D. dissertation, University of TromsЬ. Vangsnes, A. ь. (2008), Decomposing manner how in colloquial Scandinavian. Studia Linguistica, 62(1), 119­141.

2