This project deals with Romance clitic pronouns, i.e. pronominal elements that i) cannot bear primary stress and ii) may have a strong counterpart occurring in a different syntactic position.The project aims to combine evidence from different empirical domains (historical linguistics, synchronic comparison of present day vernaculars, L1/L2 acquisition, pathology) to elaborate a markedness model accounting for the organization of Romance clitic systems.First of all, I will try to single out the factors that shape clitic inventories and constrain clitic syntax. These factors can be reduced to abstract parameters, i.e. binary choices with two possible values (positive/negative) dealing with the degree of markedness of syntactic, phonological and morphological structures.Parameters having a negative value act as filters preventing the occurrence of certain configurations across languages. Languages, however, tend to avoid such configurations by means of different repair strategies (Calabrese 2005), which in turn can be ranked according to a scale of markedness. This hypothesis (namely, a parametric account based on a markedness model with repairs) may provide a better account of the number of variants displayed by grammatical systems w.r.t. cliticization.