Unsupervised Discovery of Gendered Language through Latent-Variable Modeling

Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pages 1706–1716 Florence, Italy, July 28 - August 2, 2019. c⃝2019 Association for Computational Linguistics 1706 Unsupervised Discovery of Gendered Language through Latent-Variable Modeling Alexander Hoyle@ Lawrence Wolf-SonkinS Hanna WallachZ Isabelle AugensteinP Ryan CotterellH @University College London, London, UK SDepartment of Computer Science, Johns Hopkins University, Baltimore, USA ZMicrosoft Research, New York City, USA HDepartment of Computer Science and Technology, University of Cambridge, Cambridge, UK PDepartment of Computer Science, University of Copenhagen, Copenhagen, Denmark alexander.hoyle.17@ucl.ac.uk, lawrencews@jhu.edu hanna@dirichlet.net, augenstein@di.ku.dk, rdc42@cam.ac.uk Abstract Studying the ways in which language is gen- dered has long been an area of interest in so- ciolinguistics. Studies have explored, for ex- ample, the speech of male and female charac- ters in film and the language used to describe male and female politicians. In this paper, we aim not to merely study this phenomenon qual- itatively, but instead to quantify the degree to which the language used to describe men and women is different and, moreover, different in a positive or negative way. To that end, we in- troduce a generative latent-variable model that jointly represents adjective (or verb) choice, with its sentiment, given the natural gender of a head (or dependent) noun. We find that there are significant differences between de- scriptions of male and female nouns and that these differences align with common gender stereotypes: Positive adjectives used to de- scribe women are more often related to their bodies than adjectives used to describe men. 1 Introduction Word choice is strongly influenced by gender— both that of the speaker and that of the referent (Lakoff, 1973). Even within 24 hours of birth, parents describe their daughters as beautiful, pretty, and cute far more often than their sons (Rubin et al., 1974). To date, much of the research in soci- olinguistics on gendered language has focused on laboratory studies and smaller corpora (McKee and Sherriffs, 1957; Williams and Bennett, 1975; Baker, 2005); however, more recent work has begun to fo- cus on larger-scale datasets (Pearce, 2008; Caldas- Coulthard and Moon, 2010; Baker, 2014; Norberg, 2016). These studies compare the adjectives (or beautiful lovely chaste gorgeous fertile beauteous sexy classy exquisite vivacious vibrant battered untreated barren shrewish sheltered heartbroken unmarried undernourished underweight uncomplaining nagging just sound righteous rational peaceable prodigious brave paramount reliable sinless honorable unsuitable unreliable lawless inseparable brutish idle unarmed wounded bigoted unjust brutal Male Positive Negative Female Positive Negative MISCELLANEOUS TEMPORAL SOCIAL FEELING SPATIAL QUANTITY BODY BEHAVIOR SUBSTANCE Figure 1: Adjectives, with sentiment, used to describe men and women, as represented by our model. Colors indicate the most common sense of each adjective from Tsvetkov et al. (2014); black indicates out of lexicon. Two patterns are immediately apparent: positive adjectives describing women are often related to their bodies, while positive adjectives describing men are often related to their behavior. These patterns hold generally and the differences are significant (see §4). verbs) that modify each noun in a particular gen- dered pair of nouns, …

Content truncated to 500 words