Print

Clark, Eric M.

Mathematics

2014

MS

Understanding and statistically processing underlying trends in natural human language has been an ongoing goal in Computational Social Science. This work explores trends in several languages, using expressions found on the internet, in 20th century literature, and social media. We use a Hedonometer to measure happiness in several corpora, using human ratings of emotionally charged words. Previous work has established and tested the instrument on English corpora, discovering a bias·towards positive word usage in billions of tweets, millions of books, music. lyrics, and media articles. Until now, it has remained an open question as to whether this trend is prevalent with respect to other languages. This work extends these previous analyses through a multilingual extension of the hedonometer to uncover interesting stories and underlying trends from literature and across social media.