One well-known prediction of linguistic relativity theories is the effect of a noun’s grammatical gender on its semantics; for instance, ”key” is feminine in Spanish but masculine in German and thus might be associated with feminine traits for Spanish speakers but with masculine traits for German speak- ers. Experimental and corpus evidence for these effects has been mixed. In this work, we considered a distributional se- mantics account of putative grammatical gender effects on se- mantics and tested its predictions in Spanish, German, and En- glish (control). In Part 1, we hypothesized that grammatical gender of concrete nouns affects the similarity of noun em- beddings to embeddings of adjectives semantically associated with men or with women. We found support for this hypoth- esis, showing that nouns with the same meaning but with op- posite genders in Spanish and German show opposite attrac- tion effects both for words ”man” and ”woman” and for ad- jectives associated with men and women (although the effect size was weaker for German than for Spanish). In Part 2, we asked whether people systematically choose adjectives associ- ated with women for grammatically feminine nouns and adjec- tives associated with men for grammatically masculine nouns. In a forced choice noun-adjective matching experiment (432 participants total), we found predicted grammatical gender ef- fects for Spanish but not for German. Cosine similarity be- tween the noun and the adjectives significantly predicted trial- level responses in all 3 languages; however, Spanish showed an additional effect of the noun’s grammatical gender, indicat- ing that participant noun-adjective associations are not fully explained by distributional semantics.

Authors:

George Rocco Flint: University of California, Berkeley; Anna Ivanova: Georgia Institute of Technology