ИСТИНА

It is well known that in Slavic languages, historical evolution of relative constructions (RCs) is rather uniform: the distribution of relative clauses with an internal nucleus, e.g. correlative structures, gets more restricted and gives rise to modern constructions with the external nucleus. In particular, Russian (like other languages of the area) developed an externally headed relative construction with the postnominal relative clause containing a relative pronoun kotoryj, which clearly corresponds to the interrogative pronoun of the earlier correlative structure. It may seem that this new relativization strategy is no longer connected to its diachronic source. Thus, Russian correlative constructions and postnominal externally headed relative constructions differ as to their interpretation (maximalizing vs. restrictive/appositive, Grosu&Landman 1998), available determiners (demonstrative pronouns / universal quantifiers vs. any determination) as well as to their distribution across various discourse traditions and genres (oral vs. written speech, low vs. high style). However, we argue that the postnominal externally headed relative construction with kotoryj still retains some vestiges of its origin. Specifically, we argue that it manifests a significant number of connectivity effects (Bianchi 2002), which provide us with evidence that Russian externally headed RCs are synchronically derived from configurations with the internal nucleus. Connectivity effects are morphosyntactic, selectional and interpretational effects that can only (or more naturally) be explained if we admit that the relative “head” first merges in the relativization site. In this paper, we discuss only a subset of them that are characteristic of restrictive relative clauses with kotoryj: (i) idiom relativization: the relative head can form an idiomatic expression with the internal material of the relative clause (1) (cf. Vergnaud 1974); (ii) anaphor binding: the anaphoric expressions within the relative head are bound by the relative clause internal subject (2) (cf. Schachter 1973, Sauerland 1998); (iii) scope: adjectival operators like edinstvennyj ‘only’, poslednij ‘last’, sledujuschij ‘next’, ordinal numerals and superlatives can have a “low” reading (3) (cf. Bhatt 2002). We believe that this evidence strongly suggests that the relative head in Russian restrictive RCs merges internally to the relative clause and thus supports a raising analysis of externally headed RCs with kotoryj. Interestingly, Russian RCs with interrogative pronouns (kto ‘who’, čto ‘what’), which can modify not only pronominal, but also nominal heads, never exhibit any connectivity effects. This difference between kotoryj and kto / čto RCs is consistent with their distribution in earlier constructions: kotoryj is a D head embedding an NP, whereas kto / čto are DP proforms.