ANEW+: Automatic Expansion and Validation of Affective Norms of Words Lexicons in Multiple Languages

ANEW+: Automatic Expansion and Validation of Affective Norms of Words Lexicons in Multiple Languages Samira Shaikh, Kit Cho, Tomek Strzalkowski, Laurie Feldman, John Lien, Ting Liu and George Aaron Broadwell State University of New York – University at Albany 1400 Washington Avenue Albany NY 12222 E-mail: sshaikh@albany.edu Abstract In this article we describe our method of automatically expanding an existing lexicon of words with affective valence scores. The automatic expansion process was done in English. In addition, we describe our procedure for automatically creating lexicons in languages where such resources may not previously exist. The foreign languages we discuss in this paper are Spanish, Russian and Farsi. We also describe the procedures to systematically validate our newly created resources. The main contributions of this work are: 1) A general method for expansion and creation of lexicons with scores of words on psychological constructs such as valence, arousal or dominance; and 2) a procedure for ensuring validity of the newly constructed resources. Keywords: valence lexicon; expansion; validation 1. Introduction Affective ratings of words are important to a broad spectrum of the language research community, including scholars engaged in sentiment and opinion analysis. The Affective Norms for English Words (ANEW) lexicon (Bradley and Lang, 2009) is a well-known and highly cited lexicon of words rated by human subjects along three dimensions – valence, arousal and dominance. The corpus consists of 2,477 words. The value for each word was obtained from ratings made by undergraduate students enrolled in a university. Ratings were made on a scale of 1 to 9, where a rating of 1 denoted highly negative and 9 denoted highly positive. As of December 2014, the ANEW corpus has received over 1,400 citations on Google Scholar, making it evident that it has a large impact in the scientific community. However, one major limitation of the ANEW corpus is the relatively small size. More recently, Warriner, Kuperman, and Brysbaert (2013) created a more extensive affective norms corpus, collecting ratings for approximately 14,000 words. Their normative procedure was highly similar to that used by Bradley and Lang’s (2009). One noteworthy difference, however, is that Warriner et al.’s (2013) ratings were gathered using participants recruited through Mechanical Turk. Although the Warriner et al.’s (2013) paper is an important extension of the ANEW corpus, 14,000 words may not be sufficient for researchers who are working with a large amount of text, as is often the case in fields such as natural language processing. Thus, a larger corpus was needed. Furthermore, there is currently only one such resource (Redondo et al., 2007)!for Spanish, and the number of words in this corpus is very limited (1,034); no comparable resources for affective norms of words currently exist for any other language. In this article, we first describe our method of automatically expanding an existing affective lexicon for English. Our approach is general enough to apply to any specialized lexicon in any language where a partial resource exists. We also describe a method of automatically creating lexicons for a …

Content truncated to 500 words