Aff-Lev: Developing an Affective Lexicon for Levantine neo-Arabic ​

In the diglossic Arabic-speaking world, the emergence of social media has catalyzed a transformation in language use and opinion expression, necessitating the development of suitable Natural Language Processing tools and resources. This thesis underlines the importance of capturing the nuances of Arabic vernaculars, moving away from the typical view that considers them corrupt dialects of the Standard language. The thesis focuses on Levantine Neo-Arabic, a vernacular that has taken advantage of the changes brought about by social media. The linguistic and emotional features of this variety, as well as its scarcity of resources, are highlighted. The thesis then outlines the creation of the Levantine Affective Lexicon (Aff-Lev), consisting of Levantine adjectives extracted from a corpus of tweets and annotated with a polarity score using best-worst scaling. The results of this thesis, including the novel resource Aff-Lev and the related computational linguistic analysis, provide valuable insights for individuals, organizations, and decision-makers in the Arabic-speaking world and beyond. ​