Abstract
Romance clitics, and more precisely, clitic pronouns, have been a topic in generative linguistics since the seventies. For linguists like Perlmutter 1971 and Kayne 1975, their most striking property was their linear order, which is different from the order of the phrases that realize the same grammatical functions. In more recent work, the focus has shifted from how clitics can be derived to how they are to be represented. New proposals on the treatment of Romance clitics have been made in the framework of optimality theory (Grimshaw 1997) and of feature and exponence theories, also known as Distributed Morphology (Monachesi 1999, 2000; Everett 2000; Luis and Sadler 2001).
I will argue that the classical view, according to which Romance clitics are visible elements of c-structure is adequate and that there is no need to revise the classical LFG framework in order to account for their grammatical properties. Under this assumption, two facts still need to be accounted for: the restrictions and idiosyncracies of clitic clusters; and the high ambiguity and syncretism of clitics. These are discussed here.