A computational approach to politeness with application to social factors

Proceedings of the 51st Annual Meeting of the Association for Computational Linguistics, pages 250–259, Sofia, Bulgaria, August 4-9 2013. c⃝2013 Association for Computational Linguistics A computational approach to politeness with application to social factors Cristian Danescu-Niculescu-Mizil∗‡, Moritz Sudhof†, Dan Jurafsky†, Jure Leskovec∗, and Christopher Potts† ∗Computer Science Department, †Linguistics Department ∗†Stanford University, ‡Max Planck Institute SWS cristiand|jure@cs.stanford.edu, sudhof|jurafsky|cgpotts@stanford.edu Abstract We propose a computational framework for identifying linguistic aspects of polite- ness. Our starting point is a new corpus of requests annotated for politeness, which we use to evaluate aspects of politeness theory and to uncover new interactions between politeness markers and context. These findings guide our construction of a classifier with domain-independent lexi- cal and syntactic features operationalizing key components of politeness theory, such as indirection, deference, impersonaliza- tion and modality. Our classifier achieves close to human performance and is effec- tive across domains. We use our frame- work to study the relationship between po- liteness and social power, showing that po- lite Wikipedia editors are more likely to achieve high status through elections, but, once elevated, they become less polite. We see a similar negative correlation between politeness and power on Stack Exchange, where users at the top of the reputation scale are less polite than those at the bot- tom. Finally, we apply our classifier to a preliminary analysis of politeness vari- ation by gender and community. 1 Introduction Politeness is a central force in communication, ar- guably as basic as the pressure to be truthful, in- formative, relevant, and clear (Grice, 1975; Leech, 1983; Brown and Levinson, 1978). Natural lan- guages provide numerous and diverse means for encoding politeness and, in conversation, we con- stantly make choices about where and how to use these devices. Kaplan (1999) observes that “peo- ple desire to be paid respect” and identifies hon- orifics and other politeness markers, like please, as “the coin of that payment”. In turn, polite- ness markers are intimately related to the power dynamics of social interactions and are often a decisive factor in whether those interactions go well or poorly (Gyasi Obeng, 1997; Chilton, 1990; Andersson and Pearson, 1999; Rogers and Lee- Wong, 2003; Holmes and Stubbe, 2005). The present paper develops a computational framework for identifying and characterizing po- liteness marking in requests. We focus on re- quests because they involve the speaker imposing on the addressee, making them ideal for exploring the social value of politeness strategies (Clark and Schunk, 1980; Francik and Clark, 1985). Requests also stimulate extensive use of what Brown and Levinson (1987) call negative politeness: speaker strategies for minimizing (or appearing to mini- mize) the imposition on the addressee, for exam- ple, by being indirect (Would you mind) or apolo- gizing for the imposition (I’m terribly sorry, but) (Lakoff, 1973; Lakoff, 1977; Brown and Levin- son, 1978). Our investigation is guided by a new corpus of requests annotated for politeness. The data come from two large online communities in which members frequently make requests of other mem- bers: Wikipedia, where …

Content truncated to 500 words