Cross-lingual Embeddings Reveal Universal and Lineage-Specific Patterns in Grammatical Gender Assignment

Proceedings of the 24th Conference on Computational Natural Language Learning, pages 265–275 Online, November 19-20, 2020. c⃝2020 Association for Computational Linguistics https://doi.org/10.18653/v1/P17 265 Cross-lingual Embeddings Reveal Universal and Lineage-Specific Patterns in Grammatical Gender Assignment Hartger Veeman Uppsala University Department of linguistics and philology Box 635, 75126 Uppsala hartger.veeman.7544@student.uu.se Aleksandrs Berdicevskis University of Gothenburg Spr˚akbanken Box 200, 40530 Gothenburg aleksandrs.berdicevskis@gu.se Marc Allassonni`ere-Tang University Lyon 2 Lab Dynamics of Language 14 avenue Berthelot 69363 Lyon marc.tang@univ-lyon2.fr Ali Basirat Uppsala University Department of linguistics and philology Box 635, 75126 Uppsala ali.basirat@lingfil.uu.se Abstract Grammatical gender is assigned to nouns dif- ferently in different languages. Are all fac- tors that influence gender assignment idiosyn- cratic to languages or are there any that are uni- versal? Using cross-lingual aligned word em- beddings, we perform two experiments to ad- dress these questions about language typology and human cognition. In both experiments, we predict the gender of nouns in language X using a classifier trained on the nouns of lan- guage Y, and take the classifier’s accuracy as a measure of transferability of gender systems. First, we show that for 22 Indo-European lan- guages the transferability decreases as the phy- logenetic distance increases. This correlation supports the claim that some gender assign- ment factors are idiosyncratic, and as the lan- guages diverge, the proportion of shared in- herited idiosyncrasies diminishes. Second, we show that when the classifier is trained on two Afro-Asiatic languages and tested on the same 22 Indo-European languages (or vice versa), its performance is still significantly above the chance baseline, thus showing that universal factors exist and, moreover, can be captured by word embeddings. When the classifier is tested across families and on inanimate nouns only, the performance is still above baseline, indicating that the universal factors are not lim- ited to biological sex. 1 Grammatical gender assignment Grammatical gender is one of the nominal classifi- cation systems found in natural languages (Seifart, 2010). In languages with grammatical gender, cer- tain words agree in a specific form with the noun they modify depending on the gender of the mod- ified noun (Corbett, 1991, 2001). For instance, Swedish has a binary gender system with the com- mon/neuter values, in which the articles and ad- jectives must have grammatical gender agreement with the noun they are modifying, c.f., ett stor-t ¨apple (SG.NEUT big-SG.NEUT apple.SG.NEUT) ‘a big apple’ and en stor-∅ h¨ast (a.SG.UTER big- SG.UTER horse.SG.UTER) ‘a big horse’. The most common gender distinctions are mas- culine/feminine (e.g., in French and Italian), mascu- line/feminine/neuter (e.g., in Russian and German), and common/neuter (e.g., in Swedish and Danish) (Corbett, 2013a).1 Within these distinctions, gram- matical gender does not necessarily fully agree with the biological sex. By way of illustration, the German word for ‘girl’, M¨adchen, is a neuter noun. Moreover, nouns with the same meaning may belong to different grammatical gender in dif- ferent languages. For example, the German noun for ‘sun’, Sonne, is feminine, but its French equiva- lent, soleil, is masculine. Based on these observations, several questions have …

Content truncated to 500 words