Discursive Functioning of Parts of Speech in Political Parties’ Manifestos in Pakistani Election 2013

The political discourse plays an essential role in shaping the perceptions and ideologies of people. The present research, titled ‘Discursive Functioning of Parts of Speech in Political Parties’ Manifestos’, focuses on the role of the discursive use of the parts of speech in Pakistani political parties’ manifestos, pertaining to the general election of 2013. The research is aimed at the linguistic analysis of the political discourse presented in the parties’ manifestos, focusing on parts of speech. The Van Dijk Socio Cognitive Model (1998) and Tajfel’s and Turner’s Social Identity Approach (1979) have served as the theoretical framework for this study. Corpus-based methodology was used for the Critical Discourse Analysis of the five political parties selected on the basis of popularity. The research revealed that all political parties studied used different parts of speech as a discursive strategy to frame positive images of their own party and negative images of the other parties. This was done by the parties in order to transform people’s ideology and identities in their own favor.

Variance of Maximum Likelihood Estimates for the Hidden Markov Model With Multipartite Graph Structure Transition

This study was conducted with the aim to derive an expression for variance of the maximum likelihood estimators of the hidden Markov model having multipartite graph structure transition. To obtain the estimates of variance, observed information matrix was derived using the Louise (1982) method. This study derived information matrix for the m1 and m2 partition of states which were observed at time 2t ? 1 and 2t respectively. Also, in this study, lower bound for variance of maximum likelihood estimates was derived. The study also defines a parametric bootstrap procedure for computation of variance. To check the validity of derived matrix for maximum likelihood estimates, a numerical example was used to estimate the variance using derived information matrix and compared with the results of parametric bootstrap. For this purpose, a real world data, named, as ”faithful” considered, which is freely available in statistical software R. The data-set have 272 observations on each of two variables i.e. eruption time te and waiting time tw, both measured in minutes. In this study, variable te was considered in one partition of states, which observed at time 2t?1 and variable tw was considered in second partition of states, which observed at time 2t. The study compared estimated variances by observed matrix and parametric bootstrap procedure for different combination of states and sample sizes. The comparison showed a smaller variation in values of maximum likelihood estimates obtained from observed matrix than by bootstrap procedure. In combination of states, both approaches showed almost similar variances. The overall comparison indicates that estimated variance of maximum likelihood estimators by observed matrix seems meaningful i.e. explaining less variation than that obtained from the bootstrap procedure. To study the empirical performance of the derived observed matrix for variance of maximum likelihood estimators, an extensive simulation study of various sample size was conducted. Simulated data were generated for different sizes and variance was calculated by observed matrix.