Capturing Cultural Differences in Expressions of Intentions

Proceedings of COLING 2014, the 25th International Conference on Computational Linguistics: Technical Papers, pages 48–57, Dublin, Ireland, August 23-29 2014. Capturing Cultural Differences in Expressions of Intentions Marc T. Tomlinson David B. Bracewell Language Computer Richardson TX 75080 marc, david, wayne@languagecomputer.com Wayne Krug Abstract The intersection of psychology and computational linguistics is capable of providing novel au- tomated insight into the language of everyday cognition through analysis of micro-blogs. While Twitter is often seen as banal or focused only on the who, what, when or where tweets can ac- tually serve as a source for learning about the language people use to express complex cogntive states and their cultural identity. In this contribution we introduce a novel model which cap- tures latent cultural dimensions through an individual’s expressions of intentionality. We then show how these latent cultures can be used to create a culturally-sensitive model which provides enahnced detection of signals of intentionality in tweets. Finally, we demonstrate how these models reveal interesting cross-cultural differences in the goals and motivations of individuals from different cultures. 1 Introduction Social media platforms have enabled new forms of discourse and have also provided enormous quantities of data on these communications. For instance, the popular microblogging service Twitter provides an exceptionally use- ful source of user-generated content which has attracted considerable interest from researchers in computational linguistics (Ritter et al., 2009; Gimpel et al., 2011). Most of the language processing on tweets has involved the identification of sentiment (Davidov et al., 2010), summarization (Sharifi et al., 2010), conversational mod- els of Dialogue acts (Ritter et al., 2009), or lexical and semantic processing. In this effort we expand on these previous approaches and show how individuals express their cultural identity through expressions revealing their intentionality towards events and provide a way of capturing this information. We define intentionality as the amount of effort an individual is willing to expend to achieve a goal(Ajzen, 1991). Goals represent future states or events which an individual wishes to happen. Accordingly, intentions are goals for which an individuals is willing to expend at least some minimal amount of effort to bring about. While people express goals throughout the day, intentions are the goals that they are willing to follow through with. Identifying when a goal is actually an intention requires the successful recognition of many distinct cognitive factors that can be revealed through the individual’s use of language. There is a long history of studies that have worked towards identifying a set of factors that underly an individual’s intentions (Ajzen and Fishbein, 1977; Ajzen, 1991; Malle and Knobe, 1997; Sloman et al., 2012) of which, the setting of goals is one important factor. These studies have concentrated on identifying the factors that affect an individual’s motivation. The studies have also identified a set of factors that people use to gauge the intentionality of other individuals. However, these factors have always been manually identified by an expert from an individual’s speech or writing. It is not clear that these features …

Content truncated to 500 words