Introduction

Abstract One of the basic characteristics of infinitive constructions, according to standard assumptions, is the fact that they lack an overt subject. This property has repeatedly been emphasized in generative grammar, and it has influenced the development of common theories about phenomena like empty categories, raising, and case assignment. Some regular exceptions have been noted, however, and these fitted quite well into the assumptions concerning infinitives, such as exceptional case marking (ECM) and the insertion of the preposition for in English, where the infinitival subject appears in the accusative case. However, as demonstrated by studies like Rizzi (1982) and Raposo (1987), some Romance languages permit specified subjects in a far broader range of infinitive constructions, remarkably also in the nominative case, which would definitely fall prey to the case filter in languages like English and German. For these phenomena, scholars in generative grammar have usually emphasized the language-specific properties involved in such constructions. What has not been sufficiently underlined is the fact that specified subjects in infinitival environments can be considered a common property in most Romance languages and can be found right from the first texts, appearing in the Middle Ages.