A Cross-cultural Analysis of Monologues Using Random Forest Classifier

The objective of this study is to examine how English monologues produced by Asian learners of English vary across different countries. For the analysis of the differences across countries, this study investigated whether the monologues of learners from each country exhibit characteristics of essays or dialogues regarding syntactic complexity. To determine whether the monologues are closer to essays or dialogues, a supervised machine learning technique, specifically the Random Forest Classifier, was employed and the essay and dialogue data from the International Corpus Network of Asian Learners of English were used to train the model. The Random Forest Classifier was trained to classify monologues into essays and dialogues on the basis of syntactic complexity indices. After the trained model sorted the monologues from each country into essays and dialogues, the countries were grouped based on the counts of essays and dialogues. For further analysis, Euclidean distance was utilized to identify the country whose monologues’ syntactic complexity most closely resembles that observed in Korea’s monologues, as well as the country whose monologues’ syntactic complexity most closely matches that observed in English-speaking countries’ monologues. The results demonstrated that (i) Hong Kong, the Philippines, Singapore, and the set of English-speaking countries were assigned to a group with a higher count of essays, while China, Indonesia, Japan, Korea, Pakistan, Taiwan, and Thailand were assigned to a group with a higher count of dialogues; (ii) Thailand’s monologues are most similar to those of Korea; (iii) Singapore’s monologues are most similar to those of English-speaking countries.