Automatic Extraction of Keywords in Political Speeches

The article focuses on the detection of keywords in the political speeches exemplified by the speeches of the American Senator John McCain. Harnessing keyness by way of corpus tools has become a challenging undertaking for modern computational linguistics. Relying on subtle statistical procedures leveraged to account for the salience of particular lexical items and their usage can prove to be especially beneficial while dealing with vast textual material. The UAM Corpus Tool applied in the research allowed us, based on frequency, to yield the keyness output in McCain's speeches. We have compared the keywords at the beginning and at the end of the political career and linked them with the values promoted by the politician. The conducted analysis revealed the advantages and disadvantages of the application of the corpus tool to the study of keyness. Among the advantages is time-saving text processing and the identification of subliminal keywords. Among the challenges of the methodology is overreliance on frequency counts and the need for outlining the appropriate reference corpus. The implication of the conducted computational study of the keywords is the contribution to the corpus based critical discourse analysis, pragmatics and general rhetoric of political speeches.