| Popis: |
Infinitives are trivially considered to license PRO, a covert null-Case pronoun, which represents the understood subject of an infinitive complement. However, two Romance infinitive constructions deviate from this canonical pattern and have their subject position filled by a phonetically realised lexical noun (or pronoun), which has nominative Case. In other words they license pro, which normally represents the understood subject of a finite clause. It follows that these infinitive constructions pose a problem for current theories of Control. The present thesis investigates inflected and personal infinitives in Romance languages from a synchronic and a diachronic perspective. The main goal is to offer a unified treatment of inflected and personal infinitives with regard to subject realisation, Case-licensing, Tense and non-obligatory control properties. Synchronically, on an empirical level, the generalisation is that infinitives with nominative subjects cannot surface as complements unless they bear agreement or are introduced by a complementiser. First, it is argued that agreement is not crucial to the licensing of nominative Case because other non-finite constructions, for example gerunds, license nominative subjects despite its absence. However, the latter is shown to correlate with postverbal subjects. Second, a minimalist analysis of inflected and personal infinitives is presented. It argues that both constructions are similar in that Tense licenses nominative Case. on the subjects but vary with respect to distribution: personal infinitives -unless they are introduced by a complementiser- cannot surface as complements whereas inflected infinitives can. Third, the non-obligatory control properties of inflected and personal infinitives are argued to derive from Agree not applying at CO (cf. Landau 2000), thus blocking the matching of features between the matrix and the infinitive subject, i.e. a controlled interpretation. Diachronically also, inflected and personal infinitives are intriguing since they have no direct lineage from Latin. It is proposed that there is no unique source and that inflected and personal infinitives have distinct sources. First, it is argued that the Accusativus cum Infinitivo construction cannot be proven to be the predecessor of these constructions. Second, the Spanish personal infinitive is shown to have two sources: for the personal infinitives as complements there is a learned source which has ceased to be productive and for personal infinitives as adjuncts a learnability-based account is put forward. The latter, based on a theory of acquisition and change, provides a model for the emergence of personal infinitives as adjuncts in Old Castilian that is consistent with the synchronic analysis. |
