|
|
Diachronic comparison of synchronic corpora
Křen, Michal ; Čermák, František (advisor) ; Kučera, Karel (referee) ; Králík, Jan (referee)
The thesis presents a method for diachronic comparison of synchronic corpora that reflect language of very close time periods. Its primary aim is the assessment of possi- bilities and limitations of language change detection based on the synchronic written SYN-series corpora. The approach is corpus-driven, based on a statistical evaluation of differences among normalized average reduced frequencies of lemmata and lexical combinations. There are several variants of the method applied on various subcorpora of corpus SYN and their results examined in detail. Difficulty of the comparison lies in the influence of corpus composition and the interconnection of changes in language with changes in society. As it is not easy to distinguish the signs of diachronic shift from naturally existing synchronic variability, the statistically discovered significance of frequency differences is additionally verified by querying the base corpora. The interpretation of the results is also adjusted by the knowledge of their exact composition. The conclusions are based mainly on the newspapers as a written text type that is most receptive to the changes. The changes can be characterized as a thematic diversion from the original political and economical orientation of the newspapers towards real- life and free-time topics...
|
|
|
UN Secretariat: Representative sample of the world population or dominance of specific groups?
Bartošová, Kristýna ; Parízek, Michal (advisor) ; Karlas, Jan (referee)
This thesis attempts to apply the Representative Bureaucracy theory established and developed within the Public Administration scholarship addressing the lack of formal accountability of bureaucratic bodies on the case of the largest international bureaucracy in the world - the United Nations Secretariat. It builds on the normative presumption that it is necessary to staff the bureaucracy proportionally from all the societal groups so that it reflected the values of the society in whole and the policy outcomes corresponded to those produced if all the society participated in the process. Accordingly, we aspire to find out whether the UN Secretariat is a representative sample of the world population in terms of its bureaucrats' national affiliation, and, if not, what are the factors associated with a better relative representation of a Member State in the UN Secretariat. Through the usage of descriptive statistics tools and Ordinary Least Squares Multiple Linear Regression, we find out that the per capita representation of different Member States in the UN Secretariat is by no means equal and, thus, the UN Secretariat is not a representative sample of the world population. Moreover, the research identified internal capacity of a country to ensure wellbeing and opportunities of its people and low...
|
|
|
Election Polls in Czech Republic: Methodological Optimalizations
Prokop, Daniel ; Krejčí, Jindřich (advisor) ; Soukup, Petr (referee)
Bibliographic record PROKOP, Daniel. (2012). Election polls in the Czech Republic: Methodological Optimization. Charles University in Prague, Faculty of Social Sciences, Institut of Sociological Studies. Thesis academic consultant: Mgr. Jindřich Krejčí, Ph.D. Abstract The thesis focuses on the election-polls and prediction of election results in the Czech Republic. Using data of research company MEDIAN s.r.o. from face-to-face (CAPI) and telephone interviewing (CATI) in election year 2010 it examines possibilities of methodological optimizations which could lead to reducing systematic bias and discrepancies of pre-election polls the election results. In particular, it discusses these methodological solutions: mix-mode data collection (combination of CATI and CAPI), data weighting focused on specific factors correlated with voting behavior, including preferences of undecided voters, prediction of the respondents' participation in elections, election-polls results time-series smoothing. Based on these analyses the thesis tries to articulate general findings which could be fruitful in discussion about Czech election-polls and their methodology in general. In the thesis, basic and advanced statistic methods (CART, exponential smoothing, etc.) are being used to achieve given research goals. Keywords: election...
|
|
|
Diachronic comparison of synchronic corpora
Křen, Michal ; Čermák, František (advisor) ; Kučera, Karel (referee) ; Králík, Jan (referee)
The thesis presents a method for diachronic comparison of synchronic corpora that reflect language of very close time periods. Its primary aim is the assessment of possi- bilities and limitations of language change detection based on the synchronic written SYN-series corpora. The approach is corpus-driven, based on a statistical evaluation of differences among normalized average reduced frequencies of lemmata and lexical combinations. There are several variants of the method applied on various subcorpora of corpus SYN and their results examined in detail. Difficulty of the comparison lies in the influence of corpus composition and the interconnection of changes in language with changes in society. As it is not easy to distinguish the signs of diachronic shift from naturally existing synchronic variability, the statistically discovered significance of frequency differences is additionally verified by querying the base corpora. The interpretation of the results is also adjusted by the knowledge of their exact composition. The conclusions are based mainly on the newspapers as a written text type that is most receptive to the changes. The changes can be characterized as a thematic diversion from the original political and economical orientation of the newspapers towards real- life and free-time topics...
|
guest ::
