Style of Religious Texts in 20th Century

In this study, we present the results of the investigation of diachronic stylistic changes in 20th century religious texts in two major English language varieties – British and American. We examined a total of 146 stylistic features, divided into three main feature sets: (average sentence length, Automated readability index, lexical density and lexical richness), part-of-speech frequencies and stop-words frequencies. All features were extracted from the raw text version of the corpora, using the state-of-the-art NLP tools and techniques. The results reported significant changes of various stylistic features belonging to all three aforementioned groups in the case of British English (1961–1991) and various features from the second and third group in the case of American English (1961–1992). The comparison of diachronic changes between British and American English pointed out very different trends of stylistic changes in these two language varieties. Finally, the applied machine learning classification algorithms indicated the stop-words frequencies as the most important stylistic features for diachronic classification of religious texts in British English and made no preferences between the second and third group of features in diachronic classification in American English.