Metaphorical patterns and the subprime mortgage crisis: Towards cross-linguistic, discourse-specific and n-gram-based dictionaries for sentiment analysis
In the face of international financial markets and recurrent financial crises with a global range, cross-language sentiment classification is one of the most active research areas in the field of natural language processing (NLP). In fact, this issue has spread outside of NLP to finance and accounting as well as to linguistics. So far, there is a lack of cross-linguistic resources for such analyses that allow sentiment classification in multiple languages. This paper finds that metaphorical patterns are a powerful starting point to build a cross-linguistic, discourse-specific and n-gram-based resource for sentiment classification: the large compiled corpora contain a collection of English and German newspaper articles that were published in the New York Times and in the Frankfurter Rundschau. In this paper I present the findings of a corpus-based study of metaphorical patterns used to refer to the subprime mortgage crisis of the variety of metaphors in newspaper articles, specifically regarding metaphors as a transtextual phenomenon in the light of discourse analysis. The data suggest that in both languages corresponding metaphorical patterns are used. Finally, I will discuss implications for sentiment analysis and argue for the movement towards the development of discourse-specific and n-gram-based sentiment dictionaries.