The paper deals with a morpho-syntactic innovation shared by a group of IE languages: the grammaticalisation of the pronominal linkers in verbless relative NPs as markers of nominal dependency. Three goals are pursued: 1) defining a parametrised framework for a unified description of the highly variable outcomes that this pattern has produced in the historically attested languages (for the first time all collected together); 2) critically revising the existing partial assessments of this isogloss (starting from Benveniste, 1966), in order to exclude some languages whose data are too scarce or inconsistent (e.g. Hittite, Latin), and to reassess some other data under a new viewpoint (e.g. the two competing verbless relatives in Greek); 3) suggesting some possible new members of the isogloss by applying the parametrised approach to previously unnoticed data (e.g. Khotanese, Middle Indo-Aryan, Carian). In conclusion, after a typological comparison with a parallel development in some Semitic languages, an attempt is made to describe and explain the grammatical motivations that could have initiated this phenomenon.