The Romanian Neuter Examined Through A Two-Gender N-Gram Classification System

In this paper we look at the gender system of Romanian and investigate, using machine learning techniques, the validity of the traditional analysis according to which Romanian is a three gender language. We offer strong evidence in favor of the two gender system analysis proposed in (Bateman and Polinsky, 2010) with classification accuracy higher than the one previously obtained and diverge from the approaches found in the works of (Cucerzan and Yarowsky, 2003) and (Nastase and Popescu, 2009) on automated classification of Romanian nouns according to gender, which leads us to the best accuracy in discriminating the neuter.