Modelling environmental perceptions using critical discourse analysis and philological interpretations

Abstract

In recent years, ecological degradation and environmental contamination have escalated significantly. Specialists across multiple disciplines have examined pertinent topics from numerous perspectives. Ecolinguistics arose to avert the rapid degradation of the natural setting. Critical Discourse Analysis (CDA)constitutes a significant component of ecolinguistic study, critically examining language usage through its environmental context. Initially, a tone and modality system is developed from an ecological standpoint. The present research employs the environmental theology of "equal treatment, balance, and partnership" to perform a CDA of commerce friction states, to elucidate the commonalities and distinctions in the trade friction narratives, and to uncover the environmental implications of global ecological factors within the conversation. In addition, this approach formulates a vector representation of abstract terms utilizing emotion lexicon resources and incorporates emotional polarity and part-of-speech characteristics of phrases. The word vectors are constructed into a text features a matrix, serving as the input during the Convolutional Neural Network (CNN) model. At the same time, the Back Propagation method is employed for model training. Based on the trained CNN approach, predictions are made on the unlabeled information, and the experimental CDA findings are evaluated. The findings indicate that throughout the training of Chinese and English information sets, the set used for training quality can approach 100%, while the loss rate is diminished to 0. The precision of classification for the Chinese language on the sample set achieves 82%, whereas, for the text in English, it can attain 92%, indicating favorable outcomes of the experiment. It offers an analytical framework for environmental CDA of news broadcasts regarding trade tensions. It holds significant implications for longitudinal studies on political information reporting across different camps.