Comparison of Zipf’s law in textual content and oral discourse

Abstract

Zipf’s law is a theory based on mathematics and linguistics that analyzes and quantifies how words are distributed within a text. It is possible to represent by graphs and statistical analyzes which are the terms that are repeated over so that a ranking of keywords is created. This research found, through the Zipf’s law, variations and uniformities of written academic papers and they presented orally. The oral presentations were inserted in video form on YouTube, it was possible to recover automatically the transcript of the audio. Using a Bash script, texts and transcribed presentations were quantified and organized, thereby creating tag clouds and tables with rankings, facilitating the analysis of the contents. It was possible to identify the spheres of content, identifying common words or not and, mathematically, analyze and compare what was written with what was presented in oral discourse.

Full document