A Digital Study on Public Speaking: NLP & Arguments Analysis of the First Corpus of Arabic Debates

The study of human language has evolved and developed with the emergence of new digital methods and tools that help to study and analyze huge amounts of text, speech, and other forms of the human language products. This kind of analysis helps to provide further understanding of how humans express and perceive ideas and arguments. This research studies competitive educational debates in Arabic as a structured and well-regulated example of public speaking that extensively relies on building and refuting logical arguments. This research investigates the relationship of some Natural Language Processing (NLP) features of a debate speech and its acceptance by people utilizing sentiment analysis, speech style analysis, and argumentation annotation. In addition, this study aims to examine how the people’s acceptance of a speech and the NLP features would change with respect to the speaker’s gender. In fact, no previous work has been done on analyzing Arabic debates using computational methods till today. This research utilizes several digital tools such as QATS, Rebsutate, Antconc, and LightTag in order to analyze the first corpus of competitive educational debates in Arabic. The corpus was created based on 12 recorded debates from QatarDebate international tournaments. It was found that professional debate adjudicators are totally independent in their evaluations form linguistic features and they mainly based their evaluation on good argumentation. On the other hand, beginner adjudicators did not always evaluated speeches based on good argumentation, but no correlation was found between the scores they gave and the sentiment indices of evaluated speeches. Moreover, no difference between male and female speakers was found in linguistic features nor in their debates scoring. The findings of this research open doors for further studies that relate the sentiment to the argumentation analysis of Arabic texts. It can also be utilized for training purposes that serves debaters and adjudicators.