Gender Transparency Facilitates Noun Selection in Russian * Irina A. Sekerina 1 , Patricia J. Brooks 1 and Vera Kempe 2 1 College of Staten Island and the Graduate Center of City University of New York 2 University of Stirling 1. Introduction Grammatical gender has pervasive effects on sentence processing in highly inflected languages. In Russian, it determines noun declension class, adjective, pronominal, and participle agreement, as well as past- tense verb forms. As shown by Corbett (1991), languages differ not only with respect to how many gender categories they have, but also in the extent to which semantic and morphophonological features of nouns correlate with category membership. Russian has three genders, mascu- line, feminine and neuter. Gender is highly transparent for at least 90% of nouns, with noun endings in the nominative case predictive of gender categorization. Most Russian masculine nouns tend to end in consonants, feminine nouns in –a or its allomorphs, and neuter nouns in –o or its allomorphs. In addition, there is a class of nouns ending in affricates (e.g., myš’ ‘mouse FEM ’, šalaš ‘hut MASC ’) or palatalized consonants (e.g., mebel’ ‘furniture FEM ’, korabl’ ‘ship MASC ’), which does not contain any morphophonological features providing cues to gender category mem- bership. Based on estimates from the 200 most frequent Russian nouns (Zasorina, 1977), these non-transparent nouns comprise about 10% of noun types. 1 In the present experiment, we examine whether morpho- phonologically non-transparent gender marking affects noun selection in native adult sentence comprehension. To this end, we utilize a feature of * We thank Merik Aminov and Yana Pugach for assistance with data coding. The project was supported by NSF ADVANCE Grant #0137851 awarded to Irina Sekerina. 1 Different frequency counts for Russian provide varying estimates for non-transparent nouns. For example, in the Uppsala Corpus of modern Russian texts (Lönngren, 1993), the list of the 1082 most frequent words of Russian contains only 4.2% non-transparent nouns.