MODIFIERS OF BARE NOUNS IN RUSSIAN The present talk explores the advantages of the representational theory by Bouchard (2002) for the account of nominal expressions in Russian. Bouchard assumes that in determinerless constructions the determiner is absent at any level, overt or covert. In contrast to proposals by Szabolcsi (1983), Stowell (1989) or Longobardi (1994), which take all nominal arguments to be necessarily headed by a determiner, Bouchard argues that bare NPs can function as arguments and be referring under certain semantic/ pragmatic conditions (cf. Dobrovie-Sorin & Laca 1996, Engelhardt 1999, among others). This semantics-driven approach seems particularly appealing for the analysis of determinerless Slavic languages like Russian or Polish, for whose bare NPs no concensus has been reached in literature till today. While some maintain that these languages lack determiners altogether (see Corver 1990, Zlati 1997, Lyons 1999, Boskovi 2005), others claim that a null DP is a viable option for Slavic languages (e.g. Rappaport 1998, Trugman 2004, 2007a/b, Pereltsvaig 2007). This talk starts with investigating semantic Number encoding in nominal expressions. It shows that Russian does not neatly align either with Germanic or Romance languages explored in Bouchard (2002), exhibiting mixed properties with respect to the tests proposed therein. On the surface, Russian seems to pair with English rather than French, since nouns are overtly inflected for number (in addition to other -features), and there are no overt, number-bearing articles in the language. This predicts highly limited noun omission in Russian, as in English. Yet, this prediction is falsified by the data, as shown in (1): (1) Mne ne nravjatsja zeltye rozy, ja ljublju krasnye. "I don't like yellow roses; I love red (ones).' In a similarly intriguing way, Number encoding on N is expected to preclude postnominal bare adjectives, as happens in English. This contention is also defied by Slavic data--Russian scientific terms discussed in Trugman (2007a) allow for bare postnominal adjectives: (2) zavoronok stepnoj bol'soj lark field.adj large 'a large field lark' These facts call for a modification of the representational theory, divorcing morphological number marking on Slavic nouns from semantic Number interpretation. In particular, it is argued that mandatory morphological number marking on all agreeing elements of a nominal expression, including the head noun (cf. `unavoidability of morphological marking' in Deprez 2005), may but does not automatically imply semantic Number-encoding. We show that some morphologically singular NPs, being an unmarked form, can be actually nonspecified for semantic Number, similarly to their French counterparts (see Bouchard 2003:283). This seems to be the case in modifying N1 de N2 constructions in French and N+GenP in Russian, where the nominal in genitive usually surfaces in unmarked Number form, as in (3): (3) French: a. les dents de cheval de Juliette (from Bouchard 2002, (30a)) Juliette's horse teeth Russian b. Masina poxodka baleriny (from Trugman 2003, (1c)) Masha's step ballet-dancer.GEN `Masha's ballet-dancer step' Both underlined modifiers do not refer to either some horse or a ballet-dancer, notwithstanding their singular number marking. Instead, they plainly characterize the head noun, similarly to NN compounds in English (see the translations in (3)). We further demonstrate how such a modification of Bouchard's theory allows us to accommodate some empirical facts that initialy appeared to invalidate it (see (2)). Specifically,