DPs and PPs often surface in a discontinuous manner. Standard Wh-movement extracts constituents out of DPs and PPs (1). Quantifiers may appear to the right of the DP which they modify semantically (as in (2)). According to Sportiche (1988), this construction emerges by the stranding of the quantifier when DP moves to Spec,IP. Whether "extraposition from NP" in (3) involves rightward movement depends on the status of the antisymmetry hypothesis (Kayne 1994, Chomsky 1995), but independent considerations may militate against a rightward movement explanation as well (see Culicover & Rochemont 1990). Noun incorporation also gives rise to discontinuous noun phrases, as (4) illustrates for Greenlandic. Finally, DPs and PPs may simply be 'split' in a considerable number of languages such as German, Croatian, Polish, Russian, Hungarian, Finnish, Latin, Ancient Greek, and Warlpiri, as (5) and (6) illustrate.

No abstract

No abstract

In Chomsky's Minimalist framework, word order variation reflects different movement options arising from interaction between parametrized morphological properties of functional items and invariant economy principles. In the simplest case, languages vary in whether a given movement (e.g. V-to-I-raising) feeds PF (French) or not (English). We consider more complex cases involving language-internal word order differences due to construction-specific movementGermanic Verb-Second, Last Resort Verb-fronting in Croatian -showing how they can be explained using the concept of Early Altruism latent in Chomsky's model. We further propose that (i) not only morphological, but also purely phonological properties of lexical items can trigger movement (clitics in Croatian); (ii) both finite and non-finite verbs raise to C in LF; (iii) English do-support is not a Last Resort operation: instead English 'simple' tenses are basically periphrastic, composed of a main verb and a -sometimes abstract -auxiliary.

The rich system of clitics, and the "clitic second" effect which shows up in simple main clauses, are two conspicuous features of Croatian. In previous work Cavar 1994a, Wilder andCavar 1994b), we have developed an analysis in which the clitic second effect results from the interaction between a syntactic clitic-placement rule and a phonological filter. In this paper, we are mainly concerned with more complex environments in which clitics do not appear in second, but in third position (hence the title), or somewhere further into the clause.In the work cited, we were concerned with one particular aspect of the clitic second phenomenon: the way that it interacts with verb movement. As illustrated in the paradigm (l)-(3), a verb may precede clitics in its clause only when no other constituent precedes the clitics (throughout this paper, clitics are marked in bold type):(1) a. Ivan ga je cesto citao. I. it be(3so) often read(pART) 'Ivan often read it.' b. Cesto gaje Ivan citao. often it be(3sc) I. read(PART)(2) a. *Gaje Ivan cesto citao. b. Citao gaje Ivan cesto.(3) a. *Ivan citao gaje cesto. b. *Citao Ivan gaje cesto. c. *Cesto citao gaje Ivan. d. *Citao cesto gaje Ivan.We suggested that pre-clitic placement of verbs reflects a V-movement operation in the syntax which is triggered by the same phonological filter that is responsible for part of the clitic second phenomenon, i.e. the ill-formedness of strings which have clitics in initial position (2a). In other words, this is a case where Brought to you by |