What’s in a name? In some languages, grammatical gender

Proceedings of the 2009 Conference on Empirical Methods in Natural Language Processing, pages 1368–1377, Singapore, 6-7 August 2009. c⃝2009 ACL and AFNLP What’s in a name? In some languages, grammatical gender Vivi Nastase EML Research gGmbH Heidelberg, Germany nastase@eml-research.de Marius Popescu Department of Mathematics and Computer Science University of Bucharest, Bucharest, Romania mpopescu@phobos.cs.unibuc.ro Abstract This paper presents an investigation of the relation between words and their gender in two gendered languages: German and Ro- manian. Gender is an issue that has long preoccupied linguists and baffled language learners. We verify the hypothesis that gender is dictated by the general sound patterns of a language, and that it goes beyond suffixes or word endings. Exper- imental results on German and Romanian nouns show strong support for this hypoth- esis, as gender prediction can be done with high accuracy based on the form of the words. 1 Introduction For speakers of a language whose nouns have no gender (such as modern English), making the leap to a language that does (such as German), does not come easy. With no or few rules or heuris- tics to guide him, the language learner will try to draw on the “obvious” parallel between grammat- ical and natural gender, and will be immediately baffled to learn that girl – M¨adchen – is neuter in German. Furthermore, one may refer to the same object using words with different gender: car can be called (das) Auto (neuter) or (der) Wagen (mas- culine). Imagine that after hard work, the speaker has mastered gender in German, and now wishes to proceed with a Romance language, for example Italian or Spanish. He is now confronted with the task of relearning to assign gender in these new languages, made more complex by the fact that gender does not match across languages: e.g. sun – feminine in German (die Sonne), but masculine in Spanish (el sol), Italian (il sole) and French (le soleil); moon – masculine in German (der Mond), but feminine in Spanish (la luna), Italian (la luna) and French (la lune). Gender doesn’t even match within a single language family: travel – mascu- line in Spanish (el viage) and Italian (il viaggio), but feminine in Portuguese (a viagem). Grammatical gender groups nouns in a lan- guage into distinct classes. There are languages whose nouns are grouped into more or less than three classes. English for example has none, and makes no distinction based on gender, although Old English did have three genders and some traces remain (e.g. blonde, blond). Linguists assume several sources for gender: (i) a first set of nouns which have natural gender and which have associated matching grammatical gen- der; (ii) nouns that resemble (somehow) the nouns in the first set, and acquire their grammatical gen- der through this resemblance. Italian and Roma- nian, for example, have strong and reliable phono- logical correlates (Vigliocco et al., 2004b, for Ital- ian). (Doca, 2000, for Romanian). In Romanian the majority of feminine nouns end in ˘a or e. Some rules …

Content truncated to 500 words