Split projections Summary

We began by outlining the claim made by Luigi Rizzi that in clauses which contain preposed focus/topic expressions, CP splits into a number of separate projections, viz. a Force Phrase/ForceP, a Topic Phrase/TopP and a Focus Phrase/FocP (with a Focus head being strong in finite clauses in English, but not a Topic or Force head). We pointed out that the split CP analysis of clauses raises interesting questions about the landing site of preposed wh-expressions; and we suggested that relative and exclamative wh-expressions move to the specifier position within the Force Phrase, but that interrogative wh-expressions move to the specifier position within the Focus Phrase in main-clause questions (though they move to the specifier position within the Force Phrase in complement-clause questions).We went on to examine Rizzi’s claim that split CP structures also contain a Finiteness Phrase/FinP. We noted his suggestion that clauses containing both a tropicalized and a focalized constituent are ForceP/TopP/FocP/FinP structures; clauses containing only a tropicalized (but no focalized) constituent are ForceP/TopP/FinP structures; clauses containing a focalized (but no tropicalized) constituent are ForceP/FocP/FinP structures; and clauses which contain neither a focalized nor a tropicalized constituent are simple CPs (with the relevant force and finiteness features being syncretized on a single C head). We went on to outline work by Chomsky, Larson and Hale suggesting that VPs can be split into two distinct projections – an inner VP core headed by a lexical verb and an outer vP shell headed by an affixal light verb. In particular, we looked at the syntax of ergative verbs like roll which are used both intransitively in structures like The ball rolled down the hill and transitively in structures like They rolled the ball down the hill. We argued that the verb phrase in the transitive structures comprises an inner VP core contained within an outer vP shell headed by a causative light verb with an AGENT subject, and that the light verb triggers raising of the verb roll from V to v. We argued that data relating to the distribution of various types of adverb lend support to the shell analysis, and we extended the shell analysis to transitive prepositional structures such as load the cart with hay. We presented a VP shell analysis for resultatives like turn the litmus-paper red, and double-object structures like get the teacher a present. We went on to argue that object-control structures like What decided you to take syntax? can likewise be analyzed in terms of a shell structure in which you originates as the subject of decide and what as the subject of a causative light verb; and we suggested that this analysis can be extended to other object-control predicates like persuade.We outlined Chomsky’s vP shell analysis of simple transitive structures like John read the book, and showed how such an analysis could be extended to unergatives if these are analyzed as transitive predicates which undergo object incorporation. We went on to outline a shell analysis of two-place unaccusative predicates, showing how this would account for the word order found in Belfast English imperatives such as Go you to school! We also saw how the shell analysis can handle raising structures such as They seem to me to be fine, if the EXPERIENCER to me is analyzed as occupying spec-VP, and if the verb seem raises from V to v and so comes to be positioned in front of to me. We concluded that intransitive verb phrases (like their transitive counterparts) have a shell structure in which the verb raises from V to v, and we showed that this would enable us to provide an account of locative inversion structures like Down the hill will roll the ball. We outlined Chomsky’s account of accusative case-marking, under which accusative case is assigned to a case-unvalued goal by a -complete transitive light-verb (i.e. one which has an external argument). We suggested a way of extending this analysis to ECM infinitive structures like The DA will prove the witness conclusively to have lied, arguing that the infinitive subject (the witness) raises to become the specifier of the VP headed by prove, and that the verb prove in turn raises to adjoin to a light verb which occupies the head v position of vP. We noted that this analysis amounts to claiming that the subject the witness of the infinitive complement raises up to become the object of the transitive verb prove. However, we argued that the analysis of ECM subjects are problematic in certain respects: firstly, it assumes th the ECM subject is raised to become an internal argument of the verb prove, even though it is not theta-marked by prove; secondly, it assumes that the lexical verb prove can have an [EPP] feature, when this is canonically a property of functional categories like T and C; and thirdly, it assumes that adverbs like conclusively can adjoin to intermediate projections, even though other grammatical operations (like movement) cannot target intermediate projections. We presented Bowers’s alternative analysis under which transitive verb phrases incorporate a TrP/Transitivity Phrase and thus have a tripartite vP+TrP+VP structure, with accusative objects (and ECM subjects) being case-marked via agreement with Tr and raising to spec-TrP, and the verb raising from V through Tr into v: we showed that such an analysis would allow us to suppose that adverbs only adjoin to maximal projections, and that only functional heads can have an [EPP] feature. We went on to show that if passive clauses also contain a TrP projection, we can provide a principled account of the preverbal position of passive complements in expletive structures like ‘There were several prizes awarded.’ We looked at ways in which the shell analysis could be extended to nominals, proposing an account of genitive case-marking in which a null -complete determiner assigns genitive case to a case-unvalued goal which it c-commands: however, we also noted that comparative evidence from Italian gives us reason to suppose that Num is the head which case-marks and attracts genitive expressions; and we highlighted empirical problems posed by the assumption that genitive case assignment involves agreement between a functional head (like D or Num) and a possessor.