Transformasi statistik senarai kekerapan kata dalam kajian berasaskan korpus: manifesto pilihan raya 2008

Abstract

This article examines the statistical transformation if wordlist frequency in corpus based studies. Based on a case study of the 2008 General Election manifestos of four parties BN, PKR, PAS and DAR comparisons are made between the manifestos based on raw frequencies, relative frequencies, and different techniques of frequencies visualisation. The article also compares the generation of keyword (based on frequency data) using different statistical test such as chi-square and log-likelihood, and the transformation of the frequency data using a multi variate correspondence analysis to uncover the (linguistics) relationship between the different manifesto in question. The results reveal that using different statistical procedures can lead to different conclusion with regards to the linguistics between the texts.