Exploring linguistic keywords in Nelson Mandela's speeches: A corpus-based political discourse study

A corpus-based Political discourse study


  • Sahar Javed University of management and technology Sialkot
  • Muhammad Saqib Zafar
  • Ayesha Mansoor


The key objective of this research is to investigate linguistic keywords in Nelson Mandela’s speeches and the hidden ideology behind them. For this purpose, the corpus-based methodology is used along with Van Dijk’s CDA model to analyse the corpus of 7849-word types and 106423-word tokens in the Antconc 3.5.8 version. The data analysis begins with the investigation of keywords by finding their log-likelihood or absolute frequencies. The log-likelihood of linguistic keywords is calculated by comparing frequencies in the study corpus against BNC as reference corpus through a log-likelihood calculator by (Rayson & Garside, 2000). The findings of the study divided the words with the highest log-likelihood values into four main categories i.e., speaker and related organizations 6.28%, central core or theme 57.86%, speaker and listener interaction 5.03%, and production of language according to situation 30.8%. Furthermore, the categorization is checked through Cohen’s Kappa test. The conclusion of the research highlighted the role of Mandela as an ace influencer, who impressed his listeners by using his linguistic power. The results of this research are limited to analysis of linguistic keywords in Nelson Mandela’s speeches but future researchers can further explore metaphorical and Marxist perspectives in his speeches.

Keywords: Antconc, BNC Reference Corpus, Cohen’s Kappa, Linguistic Keywords, Log-Likelihood.