Image Representation Of The National Anthems In English:A Study Based On The Text Clustering

As a symbol of the national spirit of the country,the national anthem,after officially recognized,relatively concentrated in reflecting the state will it wishes to convey.The national anthem text is usually short in length,and a thorough analysis of the national images it displays helps to promote the understanding of the target country more quickly and directly.In this study,180 English versions of the national anthems from 178 countries were collected through text cleaning,standardization,and tagging.High-frequency keywords,collocations,semantic preference,semantic prosody(sentiment complexity),and other text features co-occurring with them were extracted by corpus concordancers.Then,the stop word list and lexical semantic knowledge base resources such as Wordnet were utilized to implement the topic feature dimension reduction,and during which the keyword set originally embodied in the word form was transformed into the topic concept set emphasized on the image expression.Based on above features and the K-means method of unsupervised machine learning,the national anthem texts,according to the topic concept richness and the sentiment complexity,were finally clustered into four categories.Combining with historical and religious factors,the paper finds the different focuses on image representation and sentiment expression in these four groups:G1 FREEDOM-PURSUING countries(mainly including the United States,Germany,and Central Asian countries),G2 FAITHFUL countries(mainly including the United Kingdom,West and South Asia,Southeast Asia),G3 OPTIMISTIC countries(mainly including Spain,Australia,Latin America,and North Africa),and G4 TOUGH countries(mainly including China,Japan,Russia).In addition,case studies of the national anthems in certain countries and regions also help deepen our understanding of the characteristics of the national anthem from the following four aspects:inter-country difference of the similar topic concept use,commonality in image representation of geographically adjacent regions and countries,comparison of narrative structure and features of national anthems;and the connection between the religion and the national anthem.Most of the current studies on the image representation of the national anthems are qualitative case studies.This paper examines the large-scale national anthem texts and thus enables a panoramic study of the image representation of the national anthems.And the application of the bottom-up text feature extraction and unsupervised machine learning feature-based clustering can help get rid of preconceived mindsets and contribute to the generation of detailed interpretations.