Diverse Adversaries for Mitigating Bias in Training

Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics, pages 292–299 April 19 - 23, 2021. ©2021 Association for Computational Linguistics 292 “Talk to me with left, right, and angles”: Lexical entrainment in spoken Hebrew dialogue Andreas Weise1, Vered Silber-Varod2, Anat Lerner2,3, Julia Hirschberg4, Rivka Levitan1,5 1 Department of Computer Science, The Graduate Center, CUNY, New York, NY 10016, USA 2 Open Media and Information Lab (OMILab), The Open University of Israel, Raanana 43107, Israel 3 Mathematics and Computer Science Department, The Open University of Israel, Raanana 43107, Israel 4 Department of Computer Science, Columbia University, New York, NY 10027, USA 5 Department of Computer and Information Science, Brooklyn College, CUNY, Brooklyn, NY 11210, USA aweise@gradcenter.cuny.edu, vereds@openu.ac.il, anat@openu.ac.il, julia@cs.columbia.edu, rlevitan@brooklyn.cuny.edu Abstract It has been well-documented for several lan- guages that human interlocutors tend to adapt their linguistic productions to become more similar to each other. This behavior, known as entrainment, affects lexical choice as well, both with regard to specific words, such as re- ferring expressions, and overall style. We of- fer what we believe to be the first investiga- tion of such lexical entrainment in Hebrew. Us- ing two existing measures, we analyze Hebrew speakers interacting in a Map Task, a popular experimental setup, and find rich evidence of lexical entrainment. Analyzing speaker pairs by the combination of their genders as well as speakers by their individual gender, we find no clear pattern of differences. We do, however, find that speakers in a position of less power entrain more than those with greater power, which matches theoretical accounts. Overall, our results mostly accord with those for Amer- ican English, with a lack of entrainment on hedge words being the main difference. 1 Introduction Entrainment, also known as accommodation or alignment, is a widespread phenomenon in human interaction which leads interlocutors to adapt to each other to become more similar. It has been found for a variety of linguistic dimensions, includ- ing prosody (Levitan and Hirschberg, 2011), pho- netics (Pardo, 2006), syntax (Reitter et al., 2006), and lexical choice (Brennan and Clark, 1996). Lexical entrainment has been studied for sev- eral types of lexical choices from specific sets of words – such as referring expressions (Brennan and Clark, 1996), high-frequency words and task- related words (Rahimi et al., 2017), as well as hedge and cue phrases (Levitan et al., 2018) – to the wider linguistic style (Niederhoffer and Pen- nebaker, 2002). This motivates us to consider both specific word sets and overall language use here. Importantly, there are correlations between lexi- cal entrainment and interesting aspects of the con- versation. These include task success for both speaker pairs (Reitter and Moore, 2007; Nenkova et al., 2008) and groups (Gonzales et al., 2010; Friedberg et al., 2012), conversation flow and perceived naturalness (Nenkova et al., 2008), as well as power differences between the speakers (Danescu-Niculescu-Mizil et al., 2011). This sug- gests practical applications and has led to the devel- opment of entraining natural language generators …

Content truncated to 500 words