The syntax of complementation: On the connection between syntactic structure and selection

Date of Completion

January 1995


Language, Linguistics




This thesis explores the contribution of complementizers to the syntax and semantics of clausal complements within a theory of selection where the syntactic structure of complements is determined by the semantic properties of the matrix predicate.^ I argue that complementizers--to which I attribute the same diadic argument-structure generally associated with lexical predicates and, recently, also with Tense--, contribute to the interpretation of the entire CP as a proposition; consequently, the propositional/eventive dichotomy in the semantic component correlates with a CP/IP distinction in the syntax.^ Following this hypothesis, ECM-complements to propositional attitude predicates must have a CP projection over IP, while Control predicates like try, which select eventualities, take bare TenseP-complements. The incorporation of a ocomplementizer accounts for the basic properties and distribution of ECM-infinitives and that-less clauses, and various parametric differences regarding Control and lexical subjects in various languages are also derived.^ Assuming that want-infinitives are headed by a o sbfor$-preposition, I derive the three-way distinction between $ believe + ECM, want + infinitive$ and $want + for$ constructions concerning a wide range of syntactic phenomena (anaphora binding, passivization, $for + to$-construction, Negative Polarity licensing, etc.), from both the different argument-structure and Case properties and the morphological properties of the head involved in each construction. I argue that the incorporation of the \o-preposition to the matrix predicate creates a configuration where the infinitival subject may choose between two different Case-positions, and explore the theoretical consequences of this analysis for Economy and the notion of ``comparable derivations'.^ Regarding factive and propositional complements, they are argued to have the same internal CP-structure but to differ in the position they occupy with respect to the matrix predicate at LF. This asymmetry accounts for both the semantic difference in the presuppositional force of the complement and the behavior of propositional and factive contexts with respect to various syntactic phenomena (impossibility of ECM and that-less constructions, Tense-sequences, extraction, passivization, etc); the same analysis extends to explain the lack of derived nominals in non-factive contexts. ^